copyright by ganesh vijayaraghavan 2007

Copyright

by

Ganesh Vijayaraghavan

2007

The Dissertation Committee for Ganesh Vijayaraghavan Certifies that this is

the approved version of the following dissertation:

Synthesis and Characterization of Carbon anotube Supported

anoparticles for Catalysis

Committee:

Keith J. Stevenson, Supervisor

Jennifer S. Brodbelt

Bert D. Chandler

Arumugam Manthiram

David A. Vanden Bout



by

Ganesh Vijayaraghavan, B.Tech.; M.S.

Dissertation

Presented to the Faculty of the Graduate School of

The University of Texas at Austin

in Partial Fulfillment

of the Requirements

for the Degree of

Doctor of Philosophy

The University of Texas at Austin

December 2007

Dedication

To S.R and V.V

v

Acknowledgments

I would like to thank Keith Stevenson for the opportunity to be a part of

his research group and learn from him. I have benefitted tremendously in both

professional and personal aspects under his guidance.

I would also like to thank members of the Stevenson research group for all

their help and support. Stephen Maldonado was a huge help right from the

beginning. I’d like to thank Steve for his dedication to the CNT project and

offering sane advice when I needed it most. I wish him luck on his pursuit in

academia. Tim Smith was helpful whenever I needed help with the lab

instrumentation. I’d like to thank Jen Lyon or adding a touch of color to the lab

and insightful discussions. Ryan Williams was of great help in analyzing samples

on the SEM and the AFM. I’d also like to thank Ryan for help with labView. I’d

like to thank Ben Hahn and Lilia Kondrachova for equally insightful discussions

and help with the XPS analyses and ITO samples. I’d like to wish the best of luck

to Jaclyn Wiggins and Cori Atkinson with their projects and thank them for

unique perspectives on their respective areas of research.

I’d like to thank the undergraduate researchers at the Stevenson lab for

bringing fresh insight and energy into the lab. I’d like to thank Sawyer Croley and

vi

Joe Martini in particular for hard work and dedication to their projects. I wish

them the best in their future endeavors. In the same vein I’d like to thank Jenni

Soliz, Emily Barton, Stephen Morin and Alex Barksdale for enriching my

research experience. I’d like to thank Hugo Celio for his friendship, optimism and

help with various aspects of data acquisition and analysis.

I’m indebted to various members of the Department of Chemistry and

Biochemistry for their help and assistance. Members of the Vanden Bout, Bard,

Crooks, Brodbelt, Shear, Holcombe, Jones, Krische, Iverson and Cowley groups

were very gracious in sharing their equipment and knowledge. I’d also like to

thank members of the Manthiram group at the Department of Mechanical

Engineering and the Chandler group at Trinity University for help with

experiments.

Finally I’m indebted to my friends and family for their unconditional love

and support. I couldn’t have done this without them.

vii



Publication No._____________

Ganesh Vijayaraghavan, PhD

The University of Texas at Austin, 2007

Supervisor: Keith J. Stevenson

This dissertation describes the synthesis and characterization of nitrogen

doped carbon nanotube (NCNT) supported nanoparticles for catalysis,

specifically, the cathodic oxygen reduction reaction (ORR) in fuel cells. Strategies

for synthesis of mono- and bimetallic nanoparticle catalysts through dendrimer

based templating techniques and with the aid of metal organic chemical vapor

deposition (MOCVD) precursors and efficient assembly protocols of the catalysts

with the NCNTs are discussed in detail. Physicochemical properties of the

NCNTs and NCNT supported catalysts were characterized using a host of tools

including scanning electron microscopy, transmission electron microscopy,

Raman spectroscopy, x-ray photoelectron spectroscopy (XPS), thermo

gravimetric analysis, BET surface area and pore size analysis and electrochemical

techniques including cyclic voltammetry, chronocoulometry, chronoamperometry

viii

and rotating disk electrode voltammetry. Chapter 1 serves as a general

introduction and provides a brief overview of challenges associated with the

synthesis, characterization and utilization of graphitic carbons and graphitic

carbon supported catalysts in heterogeneous catalysis. Chapter 2 provides an

overview of the synthesis and characterization of systematically doped iron and

nickel catalyzed NCNTs in an effort to understand the effect of nitrogen doping

on ORR. Chapter 3 describes the use of NCNTs as supports for dendrimer

templated nanoparticle catalysts for ORR. A facile synthetic strategy for the

immersion based loading of catalysts onto NCNTs by spontaneous adsorption to

achieve specific catalyst loadings is explored. Chapter 4 details the loading of

monodisperse Pt, Pd and PtPd catalysts on the as synthesized NCNTs using

MOCVD precursors. The MOCVD route offers promise for direct dispersion and

activation of ORR catalysts on NCNT supports and eliminates a host of problems

associated with traditional solvent based catalyst preparation schemes. Chapter 5

details future directions on a few topics of interest including efficient

electrodeposition strategies for preparing NCNT supported catalysts, studies on

PtCu catalysts for ORR and finally prospects of using NCNT supported catalysts

in fuel cell applications.

ix

Table of Contents

List of Tables xii

List of Figures xiii

List of Schemes xviii

Chapter 1. Nitrogen Doped Nanocarbon Supported Catalysts for Heterogeneous

Catalysis

1.1 Introduction 1

1.2 References 9

Chapter 2. Synthesis and Characterization of Nitrogen Doped Carbon Nanotubes

for Oxygen Reduction

2.1 Introduction 12

2.2 Experimental 16

2.2.1 Fe Catalyzed CNT Synthesis 16

2.2.2 Ni Catalyzed CNT Synthesis 18

2.2.3 Electron Microscopy 18

2.2.4 BET Surface Area and Pore Size Analysis 19

2.2.5 Raman Spectroscopy 19

2.2.6 X-Ray Photoelectron Spectroscopy 19

2.2.7 Thermo Gravimetric Analysis 20

2.2.8 Electrochemical Analysis 20

2.3 Results and Discussion 22

2.3.1 CNT Growth on Substrates 22

2.3.2 TEM Observations of the Fe Catalyst and CNT Crystallinity 24

2.3.3 Surface Area and Porosity of CNTs 26

2.3.4 Raman Analysis 29

2.3.4.1 Raman Analysis at Fe Catalyzed CNTs 29

2.3.4.2 Raman Analysis at Ni Catalyzed CNTs 29

2.3.5 XPS Analysis 31

2.3.5.1 XPS Analysis at Fe Catalyzed CNTs 31

2.3.5.2 XPS Analysis at Ni Catalyzed CNTs 31

2.3.6 Effect of Increasing Fe Content in CNTs on ORR 37

2.3.7 Electrochemical Analysis at CNTs 39

2.3.7.1 ORR at Fe Catalyzed CNTs 39

2.3.7.2 ORR at Ni Catalyzed CNTs 43

x

2.3.7.3 Effect of Pyridinic Nitrogen on ORR 45

2.4 Conclusions 51

2.5 References 52

Chapter 3. Synergistic Assembly of Dendrimer Templated Catalysts and Nitrogen

Doped Carbon Nanotube electrodes for Oxygen Reduction 55

3.1 Introduction 55

3.2 Experimental 60

3.2.1 Synthesis of Monometallic Dendrimer Encapsulated NPs 60

3.2.2 Synthesis of Bimetallic Dendrimer Encapsulated NPs 61

3.2.3 Synthesis of Undoped CNTs and NCNTs 62

3.2.4 Adsorption of Pt-DENs on CNTs and NCNTs 62


3.2.6 Thermo Gravimetric Analysis 64

3.2.7 Electrochemistry at CNT/ DEN Composites 64


3.3.1 Structure and Composition of DENs 66

3.3.2 Analysis of Pt-DEN Adsorption at CNTs and NCNTs 66

3.3.3 TEM Analysis of DENs Adsorbed on NCNTs 76

3.3.4 Comparison of Adsorption Characteristics of dendrimers 78

3.3.5 TGA Analysis of Pt-DEN Loading on NCNTs 81

3.3.6 Electrochemical Analysis of CNT/ DEN Composites 83

3.3.6.1 CV Analysis at CNT/ Pt-DEN Composites 83

3.3.6.2 Rotating Disk Electrode Studies 84

3.4 Conclusions 89

3.5 References 91

Chapter 4. Metal Organic Chemical Vapor Deposition of Nanocarbon Supported

Mono- and Bimetallic Catalysts for Oxygen reduction 94

4.1 Introduction 94

4.2 Experimental 99

4.2.1 Synthesis of Undoped CNTs and NCNTs 99

4.2.2 CVD of Mono- and Bimetallic Nanoparticles on CNTs 99


4.2.4 Raman Characterization 100

4.2.5 XPS Characterization 101

4.2.6 TGA 101



4.3.1 Morphology of Catalysts Synthesized by MOCVD 103

xi

4.3.2 Raman Analysis 107

4.3.3 XPS Analysis 107

4.3.4 TGA Analysis 110


4.3.5.1 Cyclic Voltammetry 113

4.3.5.2 ESA Analysis by CO Stripping 115

4.3.5.3 Chronoamperometric Studies 117

4.3.5.4 RDE Studies 119

4.3.5.5 Stability of PtPd Catalysts 121

4.4 Conclusions 126

4.5 References 127

Chapter 5. Future Directions 130

5.1 Introduction 130


5.2.1 Electrodeposition of Nanoparticle Catalysts on NCNTs 132

5.2.2 Synthesis and Characterization of NCNT Supported PtCu 139

5.2.2.1 Studies on Dealloying at Au Thin Films 140

5.2.2.2 Effects of Electrochemical Dealloying at NCNT/PtCu 143

5.2.2.3 Development of New Characterization Strategies 145

5.2.3 CNT Supported Catalysts in Fuel Cells 146

5.4 Conclusions 148

5.4 References 148

Vita.......................................................................................................................150

xii

List of Tables.

Table 2.1 Comparison of parameters derived from Raman, XPS, cyclic

voltammetry, surface area analysis and TGA analysis for Fe and Ni

catalyzed NCNTs 50

Table 3.1 Comparison of Pt-DEN adsorption parameters, BET surface area,

electroactive surface area and peak potential for ORR at undoped

CNT/ Pt-DEN, and NCNT/ Pt-DEN composites for 4 % NCNTs and

7.5 % NCNTs 86

Table 4.1 Comparison of BET surface areas and edge plane content for pristine

nanocarbons calculated from Raman data for undoped CNTs, 4 %

NCNTs and 6.5 % NCNTs. Also compared are catalyst loadings on

the nanocarbons and ESAs obtained from integrated charge

associated with the oxidation of CO and hydrogen adsorption/

desorption for supported PtPd catalysts 118

xiii

List of Figures.

Figure 1.1 Representation of disparity in classic and new carbons. Plot shows

comparison of surface areas between the two classes of carbons 2

Figure 2.1 Schematic of the apparatus used to synthesize undoped and nitrogen

doped CNTs 17

Figure 2.2 Representative SEM image of CNTs grown directly on a Ni mesh

current collector. Image shows that CNTs retain their alignment even

after acid dissolution of the Ni 23

Figure 2.3 Representative TEM images of residual Fe catalyst particles encased

within the graphitic sheets of CNTs. a) Encapsulated Fe in undoped

CNTs. b) Fe encapsulated within 5 % NCNTs 25

Figure 2.4 Plot of differential surface area corresponding to pore sizes at

undoped CNTs and NCNTs measured using nitrogen 28

Figure 2.5 Raman spectra showing the D and G bands on nitrogen doped Ni

catalyzed NCNTs 30

Figure 2.6 Plot of ID / IG vs. the pyridinic concentration at Ni based NCNTs.

Also shown is the length of crystallinity (La) correlating degree of

structural disorder to the pyridinic fraction of surface nitrogen 32

Figure 2.7 XPS spectra showing the nitrogen region in nitrogen doped Ni

catalyzed NCNTs 34

Figure 2.8 Quantitative comparison of the different nitrogen fractions in

nitrogen doped Ni catalyzed NCNTs based on XPS spectra 36

Figure 2.9 Oxygen reduction at 4 % NCNTs with varying concentrations of Fe

added to the electrolyte solution 38

Figure 2.10 Plot showing correlation between peak potential for ORR activity

corresponding to mass of iron precursor used to synthesize nitrogen

doped CNTs 40

xiv

Figure 2.11 ORR activity for undoped CNTs and NCNTs in oxygen saturated 0.1

M H2SO4. ν = 20 mVs-1

42

Figure 2.12 Comparison of cyclic voltammograms for ORR at undoped CNTs

and NCNTs synthesized using Fe and Ni precursors in oxygen

saturated 0.1 M H2SO4. ν = 20 mVs-1

44


corresponding to edge plane content calculated from Raman spectra

at Fe and Ni based NCNTs. ORR activity was measured in oxygen

saturated 0.1 M H2SO4 46


corresponding to the pyridinic nitrogen fraction at Fe and Ni based

NCNTs. ORR activity was measured in oxygen saturated 0.1 M

H2SO4 47


corresponding to the total nitrogen content at Fe and Ni based

NCNTs. ORR activity was measured in oxygen saturated 0.1 M

H2SO4 48

Figure 3.1 Representative TEM image of Pt –DENs. The histogram depicts a

typical particle size distribution for the Pt-DENs 67

Figure 3.2 Representative TEM image of Pd –DENs. The histogram depicts a

typical particle size distribution for the Pd-DENs 68

Figure 3.3 Representative TEM image of PdAu –DENs. The histogram depicts

a typical particle size distribution for the PdAu-DENs 69

Figure 3.4 Representative energy dispersive spectrum for PdAu- DENs 70

Figure 3.5 UV-Vis spectra showing cumulative adsorption measurements

observed over 24 hours for the adsorption of Pt-DEN nanoparticles

on 4 % NCNTs 71

Figure 3.6 Picture representing the Pt-DEN adsorption process on NCNTs. A)

20 µM Pt-DENs. B) Pt-DENs with NCNTs suspended in solution. C)

NCNT/ Pt-DEN suspension after 24 hrs showing all DENs having

been adsorbed onto the NCNTs rendering the solution colorless 73

xv

Figure 3.7 Adsorption isotherms for G4-NH2 Pt-DEN adsorption on undoped

CNT and NCNT supports 74

Figure 3.8 TEM image of G4-NH2 Pt-DENs adsorbed on 4 % NCNT supports

(Scale bar is 20nm). The inset shows high resolution structure of Pt

nanoparticles (Scale bar is 5nm) 77

Figure 3.9 TEM images of dendrimer encapsulated nanoparticles adsorbed on

the NCNT surface. a) Pd DENs adsorbed on 4 % NCNTs. b) PdAu

DENs adsorbed on 4 % NCNTs 79

Figure 3.10 Representative TEM images comparing the adsorption of –NH2

terminated and –OH terminated Pt DENs on 4 % NCNTs. a) G4-

NH2 Pt DENs adsorbed on 4 % NCNTs. b) G4-OH Pt DENs

adsorbed on 4 % NCNTs 80

Figure 3.11 Representative TGA heating curves for blank NCNTs and Pt-DENs

adsorbed on 7.5 % NCNTs. Mass loading of Pt on NCNTs is

calculated by subtracting final wt. % of NCNT/ Pt-DEN from final

wt. % at blank NCNT 82

Figure 3.12 Cyclic voltammograms for ORR at control undoped CNTs and

NCNTs synthesized using Fe and Ni precursors compared to NCNT/

Pt-DEN composites in oxygen saturated 0.1 M H2SO4. ν = 20

mVs-1

85

Figure 3.13 Polarization curves for the ORR on G4-NH2 Pt-DEN/CNT and

NCNT composites supported on a glassy carbon electrode immersed

in an O2 saturated 0.1M H2SO4 solution. In all cases the Pt loading is

18±1 µg. Also shown are polarization curves for CNT and NCNT

supports. Rotation rate =1600 rpm, scan rate =20 mV s-1

87

Figure 4.1 TEM images of a) Pt and c) Pd catalysts on NCNT supports.

Corresponding particle size histograms for b) Platinum and d)

Palladium catalysts 105

Figure 4.2 TEM image of a) PtPd particles on 6.5 % NCNT supports. Inset

shows high resolution image of a PtPd bimetallic catalyst particle. b)

Corresponding particle size histogram for PtPd catalysts. c)

xvi

Representative energy dispersive spectrum of a single PtPd

nanoparticles 106

Figure 4.3 Raman spectra of PtPd bimetallic catalysts on NCNT supports with

varying surface concentrations of nitrogen 108

Figure 4.4 XPS spectra of PtPd supported on 6.5 % NCNTs. a) Survey

spectrum of PtPd showing Pt and Pd regions. b) High resolution XPS

spectrum of Pt and c) Pd catalysts supported on 6.5 % NCNTs 109

Figure 4.5 TGA heating curves for PtPd catalysts on various CNT supports 111

Figure 4.6 Effect of edge plane content at NCNTs on PtPd loading as

determined by TGA 112

Figure 4.7 Representative cyclic voltammogram for PtPd catalysts supported on

6.5 % NCNTs in O2 saturated 0.1M H2SO4 114

Figure 4.8 CO stripping voltammogram on PtPd catalysts supported on 6.5 %

NCNTs. CO was dosed into solution for 20 min followed by 30 min

of Ar purge to remove CO in bulk 116

Figure 4.9 CO stripping transients on PtPd catalysts supported on 6.5 %

NCNTs. CO was dosed into solution for 20 min followed by 30 min

of Ar purge to remove CO in bulk. The working electrode was held

at 0.11V during CO dosing and raised to stripping potentials 120

Figure 4.10 Polarization curves for ORR on PtPd catalysts supported on 6.5 %

NCNTs in O2 saturated 0.1 M HClO4. The RDE was rotated between

250 – 3000 rpm in increments of 250 rpm. Scan rates were set at 20

mV /s 122

Figure 4.11 Polarization curves for ORR on PtPd catalysts supported on 6.5 %

NCNTs in O2 saturated 0.1 M H2SO4 showing the effects of

adsorbed bisulfate. The RDE was rotated between 250 – 3000 rpm in

increments of 250 rpm. Scan rates were 20 mV /s 123

Figure 4.12 Extended cycling of PtPd catalysts supported on 6.5 % NCNTs in O2

saturated 0.1 M H2SO4 124

xvii

Figure 5.1 Chronocoulometric profile of the electrodeposition of Au

nanoparticles from 0.2 mM HAuCl4 solutions on NCNTs 134

Figure 5.2 Representative TEM images of nanoparticle catalysts deposited on 4

% NCNTs. a) Au nanoparticles. b) PtAu nanoparticles 135

Figure 5.3 High resolution TEM image of PtAu nanoparticles electrodeposited

on NCNTs 137

Figure 5.4 Cyclic voltammograms for oxygen reduction at PtAu catalysts

deposited on NCNTs. CV’s were run in O2 saturated 1 M KNO3 138

Figure 5.5 Scanning electron microscope image of Au film subjected to 30

alloy/ dealloy cycles in a ZnCl2 / benzyl alcohol solution 141

Figure 5.6 Cyclic voltammograms of an Au film in 0.5 M H2SO4 before and

after alloying/ dealloying cycles in a ZnCl2 / benzyl alcohol solution.

Integration of the cathodic AuO peak showed a 17.1 % increase in

the electrochemical surface area on the dealloyed Au film 142

Figure 5.7 Electrochemical dissolution of Cu from PtAu bimetallic catalysts

synthesized using dendrimer templates. The PtCu/ DEN catalysts

were supported on 6.5 % NCNTs. Dissolution studies were carried

out in 0.1 M HClO4 144

Figure 5.8 Half cell trial of CNT supported Pt DEN catalysts for ORR in 1 M

H2SO4 147

xviii

List of Schemes.

Scheme 3.1 Preparation of amine terminated PAMAM dendrimer encapsulated

Pt nanoparticles (G4-NH2 Pt-DENs) and adsorption onto carbon

nanotube (CNT) supports 58

Scheme 3.2 Representation of synergistic ORR activity at NCNT/ Pt-DEN

composites 90

1

CHAPTER 1

itrogen Doped anocarbon Supported Catalysts for Heterogeneous Catalysis

1.1 ITRODUCTIO

The use of carbon in a chemical process dates back as far as 3750 BC in

the Bronze Age. Early Egyptians and Sumerians were known to have used

charcoal to reduce copper, tin and zinc ores for producing bronze. Since then

carbons have evolved for use in a multitude of processes. Traditional carbons1,2 or

‘classic carbons’ such as soot, charcoal and diamond have been mostly replaced

with new and novel carbon materials or ‘new carbons’.3 Since the 1960’s a variety

of synthetic strategies have been employed to produce carbons that have favorable

structure, texture and composition that allow for these new carbons to perform

significantly better than classic carbons.4 Early examples of these new carbons

included carbon fibers, porous carbons and pyrolitic graphite electrodes.3 Of these

new carbons, graphitic carbons have shown promise for usage in energy storage

and conversion processes due to their high electronic conductivity, resistance to

corrosion and physical stability.5,6,7

Diverse synthetic protocols have been utilized in the preparation of

graphitic carbons. Fullerenes,8 are one form of graphitic carbons that were

discovered in 1985 and have been synthesized by the evaporation of carbon using

an electric arc between graphite electrodes in an inert atmosphere. Carbon

nanotubes (CNTs) discovered by Ijima9 in 1991 are another form of graphitic

2

1 2 3 4 5 6 7 80

200

400

600

800

1000

1200

Su

rfa

ce

are

a

(m2 /

g)

Carbons

1. Graphite

2. Thermax

3. Phil black A

4. Carbon black

5. PAN fibers

6. Vulcan XC 72

7. Kejtenblack

8. Functionalized CNTs

Classic carbons

New carbons

Figure 1.1 Representation of disparity in classic and new carbons. Plot shows comparison of surface areas between the two classes of carbons.

3

carbons that have been synthesized by various routes. CNTs have been produced

using templated techniques, some of which involve the deposition of carbon from

propylene gas at 800 0C on the inner walls of nanoscale channels in an aluminum

oxide film.10,11

The CNTs were then recovered by dissolution of the film in HF.

Other techniques include CNTs produced using a polymer blend process,12,13

CNTs produced with the help of a steep thermal gradient in alcohols,14 CNTs

produced by the decomposition of SiC in a single crystal wafer15 and the more

popular laser ablation technique.16,17

The popularity of CNTs stems from favorable properties such as good

electronic conductivity, surface area, preferred pore structure, ability to be

synthesized in ordered arrays,18 ease of assembly on a secondary substrate

through Van der Waal’s interactions,19,20

facility for directed growth21 – typically

brought upon by application of an electric field and scalability of the patterned

growth.22

While these as-synthesized CNTs perform well in their primary purpose as

electron conductors in chemical and electrochemical processes, it would be of

tremendous advantage if this class of graphitic carbons were modified to be active

in these chemical and electrochemical reactions themselves. Efforts towards this

end have involved heteroatom doping – most notably nitrogen doping on graphitic

carbons. A number of groups such as Thrower et al,23,24

Wang et al25, Boehm et

al,26,27

Gooding et al,28,29

Compton et al,30,31

Cai et al32 and Dodelet et al

25,33 have

studied the effects of nitrogen doping in graphitic carbons especially on the

4

oxygen reduction reaction (ORR). ORR is a technologically important reaction

that takes place in the cathodic side of a fuel cell.

A number of synthetic strategies have been employed in the nitrogenation

of graphitic carbons. The most popular method being the introduction of the as-

prepared carbon into an atmosphere of NH3 or HCN at high temperatures.25 The

resulting carbon was found to be enriched in nitrogen and had a significantly

higher activity for electrochemical reduction of oxygen. While it has been of

general consensus that nitrogen doping increases the activity at these carbons,

there has been ambiguity as to what physicochemical factor or factors effect this

difference in activity. Most of the ambiguity stems from poor reproducibility33

and stability problems associated with the post- synthetic modification step at

these carbons. To this end, effort in the Stevenson research lab has been

concentrated on developing facile synthetic protocols34,35,36

that enable synthesis

of systematically doped graphitic nanocarbons in order to elucidate individual

physicochemical properties and their contribution to ORR activity.

Chemical vapor deposition (CVD) is employed for the controlled

synthesis of nitrogen doped carbon nanotubes (NCNTs) at the Stevenson research

lab. The use of synthetic precursors in CVD synthesis allows for systematic

manipulation of structure, composition, surface area and the degree of structural

disorder at NCNTs. This systematic manipulation allows for distinctive

delineation of the effects brought upon by nitrogen doping on the electrochemical

properties of the NCNTs. Accordingly, Maldonado et al have published a series of

5

papers34,35,36

elucidating the influence of systematic nitrogen doping at NCNTs on

several electrochemical processes including ORR.

While a systematic increase in nitrogen doping at graphitic carbons

increases the activity for ORR at these nanocarbons, their activities are lower than

what is seen at metals known for exhibiting high ORR activity. Efforts undertaken

to improve on this particular aspect have involved using these carbons as supports

for active metal nanoparticle catalysts to be used in heterogeneous catalysis.

Conventional strategies employed for loading metal catalysts on carbons

typically involve inducing surface functionalities such as carbonyl, carboxylate,

ester-like oxygen or alcohol on the carbon to facilitate anchoring of catalysts on

the support. This is done using aggressive protocols that involve the use of strong

acids37 like HNO3, H2SO4, HCN or strong oxidizing agents

38 like H2O2, KMnO4.

These protocols are often not reproducible and degrade the preferred structural

and compositional properties of both the carbon support and active metal catalyst.

The ability to manipulate in-situ, the structure, composition, surface area

and the density of edge plane sites on NCNTs provides a significant advantage

over traditional carbons for use as catalyst supports because there is no further

need to modify the surface to facilitate catalyst loading. This offers tremendous

advantages in using NCNTs as catalyst supports while circumventing the time

consuming and often structurally degrading pre- synthesis steps that are used in

catalyst loading protocols like microemulsion,39 impregnation,

40 co-

precipitation,41 sonochemical

42 and fluidized bed CVD processes.

43 NCNTs as

catalyst supports also allow for uniform catalyst dispersion and utilization

6

stemming from control of their surface properties such as the density of edge

plane sites and porosity.

Challenges in defining physicochemical properties of nitrogenated

graphitic carbons (esp. NCNTs) that contribute to oxygen reduction and the

advantages in using these properties efficiently and synergistically with active

metal catalysts for ORR are discussed in this dissertation.

This dissertation is organized into five chapters. Chapter 1 serves as a

general introduction and provides a brief overview of challenges associated with

the synthesis, characterization and utilization of graphitic carbons and graphitic

carbon supported catalysts in heterogeneous catalysis.

Chapter 2 provides an overview of the synthesis and characterization of

systematically doped iron and nickel catalyzed NCNTs in an effort to understand

the effects of nitrogen doping on ORR. The influence of nitrogen doping on the

physicochemical and electrochemical properties at these NCNTs is characterized

using transmission electron microscopy (TEM), scanning electron microscopy

(SEM), Raman spectroscopy, x-ray photoelectron spectroscopy (XPS), thermo

gravimetric analysis (TGA), BET surface area and pore size analyses and

electrochemical characterization including cyclic voltammetry (CV) and rotating

disk electrode voltammetry (RDE). Raman analysis in conjunction with XPS and

electrochemical analysis suggests that the presence of increased edge plane sites

at NCNTs along with increased surface concentrations of pyridinic nitrogen are

influencing factors for remarkable ORR activity at NCNTs. BET surface area and

pore size analyses indicate that the porosity at NCNTs does not correlate to ORR

7

activity. ORR activity studies at Ni based NCNTs suggest that an Fe based active

center does not provide for a complete picture of the ORR mechanism.

Chapter 3 describes the use of CNTs as supports for dendrimer templated

catalysts (DENs) for ORR. A facile synthetic strategy for the immersion based

loading of catalysts onto a carbon support by spontaneous adsorption to achieve

specific catalyst loadings is explored. The resultant CNT-DEN composites were

assembled without necessitation of time consuming and irreproducible pre- and

post- synthetic processing steps. The advantage of this synthetic strategy is that

the compositional and structural properties of both the carbon support and the

DEN catalyst can be reproducibly prepared and synergistically tuned to directly

assemble carbon-supported catalyst composites via adsorption from aqueous

solution. The as loaded Pt-DENs were found to be uniformly dispersed

throughout the CNTs and CNT/ Pt-DEN composites were found to be active for

ORR. A synergistic activity is envisioned where the NCNT support is reactive

and serves to reduce the peroxide formed as a byproduct during oxygen reduction

at the metal catalyst.

Chapter 4 details the loading of monodisperse Pt, Pd and PtPd catalysts on

as synthesized NCNTs using metal organic chemical vapor deposition (MOCVD)

precursors. The MOCVD route offers promise for the direct dispersion and

activation of mono- and multimetallic ORR catalysts on NCNT supports and

eliminates the current inevitable problems involving loading, sintering and

activation steps associated with traditional solvent based catalyst preparation

schemes. The tunability of the amount of edge plane content on the NCNT

8

support offers distinct advantages with regard to preferential catalyst loading and

decomposition of peroxide specific to ORR. The MOCVD process is expected to

serve as a model system for rapid and direct assembly of carbon supports and

varied catalyst compositions that provide promising activity for oxygen reduction.

Chapter 5 details future directions in studies on a few topics of interest

involving efficient synthetic strategies for NCNT/ nanoparticle catalysts and their

application towards ORR. In this regard, electrodeposition aspects of loading

mono- and bimetallic catalysts directly on as synthesized NCNT electrodes is

reported. Facile characterization protocols that allow for TEM observations of the

as synthesized NCNT/ catalyst composite and the composite after being subject to

ORR are also discussed. These characterization protocols show marked

improvements to existing strategies and are expected to play a vital role in the

design of effective ORR catalysts. Other future directions discussed include

studies on PtCu catalysts for ORR. The selective dissolution of Cu from a

bimetallic PtCu catalyst presents intriguing possibilities as to a more active

catalyst surface for ORR. The electrochemical dealloying of Cu is speculated to

improve the electrochemical surface area and enhance mass transport.

Characterization of such interfaces will help design stable and active catalyst

combinations for ORR.

9

1.2 REFERECES

1. Inagaki, M. Old but new materials: carbons. In Carbons- control of structure and functions; Elsevier, 2000; p. 1 – 29.

2. Walker Jr. P. L. Carbon, 1990, 28, 261.

3. Inagaki, M.; Radovic, L. R. Carbon, 2002, 40, 2263.

4. Inagaki, M. In ew carbons – control of structure and functions; Inagaki, M., Ed.; Elsevier, 2000; p. 82-123.

5. Sato, K.; Noguchi, M.; Demachi, A.; Oki, N.; Endo, M. Science, 1994, 264, 556.

6. Endo, M.; Maeda, T.; Takeda, T., Kim, Y. J.; Koshiba, K.; Hara, H.; Dresselhaus, M. S. J. Electrochem. Soc. 2001, 148, A910.

7. Endo, M.; Hayashi, T.; Hing, S. H.; Enoki, T.; Dresselhaus, M. S. J. Appl. Phys. 2001, 90, 5670.

8. Kroto, H. W.; Heath, J. R.; O’ Brien, S. C.; Curl, R. F.; Smalley, R. E. ature, 1985, 318, 162.

9. Ijima, S. ature, 1991, 354, 56.

10. Kyotani, T.; Tsai, L.; Tomita, A. Chem. Mater. 1995, 7, 1427.

11. Kyotani, T.; Tsai, L.; Tomita, A. Chem. Mater. 1996, 8, 2109.

12. Oya, A.; Kasahara, N. Carbon, 2000, 38, 1141.

13. Hulicova, D.; Sato, F.; Okabe, K.; Koishi, M.; Oya, A. Carbon, 2001, 39, 1438.

14. Zhang, Y.; Nishitani- Gamo, M.; Nagakawa, K.; Ando, T. J. Mater. Res. 2002, 17, 2457.

15. Kusunoki, M.; Rokkaku, M.; Suzuki, T. Appl. Phys. Lett. 1997, 18, 2620.

16. Kocabas, C.; Meitl, M. A.; Gaur, A.; Shim, M.; Rogers, J. A. ano Lett. 2004, 4, 2421.

10

17. Krungleviciute, V.; Heroux, V.; Migone, A. D.; Kingston, C. T.; Simard, B. J. Phys. Chem. B 2005, 109, 9317.

18. Fan, S.; Chapline, M.; Franklin, N.; Tombler, T.; Cassell, A.; Dai, H. Science, 1999, 283, 512.

19. Cassell, A.; Franklin, N.; Tombler, T.; Chan, E.; Han, J.; Dai, H. J. Am. Chem. Soc. 1999, 121, 7975.

20. Franklin, N.; Dai, H. Adv. Mater. 2000, 12, 890.

21. Zhang, Y.; Chan, A.; Cao, J.; Wang, Q.; Kim, W.; Dai, H. Appl. Phys. Lett. 2001, 79, 3155.

22. Franklin, N.; Li, Y.; Chen, R. J.; Javey, A.; Dai, H. Appl. Phys. Lett. 2001, 79, 4571.

23. Strelko, V. V.; Kuts, V. S.; Thrower, P. A. Carbon, 2000, 38, 1499.

24. Thrower, P. A.; Radovic, L. R.; Carbon, 2000, 39, 1.

25. Wang, H.; Cote, R.; Faubert, G.; Guay, D.; Dodelet, J. P. J. Phys. Chem. B 1999, 103, 2042.

26. Boehm, H. P.; Mair, G.; Stoehr, T.; DeRincon, A. R.; Tereczki, B. Fuel 1984, 63, 1061.

27. Stohr, B.; Boehm, H. P.; Schlogl, R. Carbon 1991, 29, 707.

28. Gooding, J. J. Electrochimica Acta 2005, 50, 3049.

29. Chou, A.; Bocking, T.; Singh, N. K.; Gooding, J. J. Chem. Commun. 2005, 7, 842.

30. Banks, C. E.; Moore, R. R.; Davies, T. J.; Compton, R. G. Chem. Commun. 2004, 16, 1804.

31. Salimi, A.; Banks, C. E.; Compton, R. G. Analyst, 2004, 129, 225.

32. Jaouen, F.; Lefevre, M.; Dodelet, J. P.; Cai, M. J. Phys. Chem. B 2006, 110, 5553.

33. Jouen, F.; Charreteur, F.; Dodelet, J. P. J. Electrochem. Soc. 2006, 153, A689.

11

34. Maldonado, S.; Stevenson, K. J. J. Phys. Chem. B 2004, 108, 11375.


36. Maldonado, S.; Morin, S.; Stevenson, K. J. Carbon 2006, 44, 1429.

37. Wang, J.; Musameh, M.; Lin, Y. J. Am. Chem. Soc. 2003, 125, 2408.

38. Tian, Z. Q.; Jiang, S. P.; Liang, Y. M.; Shen, P. K. J. Phys. Chem. B. 2006, 110, 5343.

39. Yoon, B.; Wai, C. M. J. Am. Chem. Soc. 2005, 127, 17174.

40. Gaur, V.; Sharma, A.; Verma, N. Carbon 2005, 42, 3041.

41. Li, X.; Hsing, I. -M. Electrochim. Acta 2006, 51, 5250.

42. Xing, Y. J. Phys. Chem. B 2004, 108, 19255.

43. Serp, P.; Hierso, J. C.; Feurer, R.; Kihn, Y.; Kalck, P.; Faria, J. L.; Aksoylu, A. E.; Pacheco, A. M.; Figueiredo, J. L. Carbon 1999, 37, 527.

12

CHAPTER 2

Synthesis and Characterization of itrogen Doped Carbon anotubes for Oxygen Reduction

2.1 ITRODUCTIO

Graphitic carbons have been used as supports for a variety of processes

and especially in heterogeneous catalysis.1-4 The interest in using these carbons

stems from their good electrical conductivity,5,6 high surface area7 and good

mechanical and chemical stability.8 While these carbons perform well in their

primary purpose as electron conductors for chemical and electrochemical

processes, it would be of tremendous advantage if these carbons were tuned to

take part in these chemical and electrochemical reactions themselves. This

scenario would make more efficient use of the surface area of these supports in

devices like fuel cells and batteries.

One technologically important reaction where graphitic carbons are used

as catalyst supports is the oxygen reduction reaction (ORR) that takes place in the

cathodic side of a fuel cell.9-11 Efforts to enhance the oxygen reduction activity at

these carbons have been afoot since the late 60’s12 have increased substantially

since the early 80’s partly due to the re-emergence of fuel cells as promising

‘green’ power sources.13 Heteroatom doping of these graphitic carbons especially

with nitrogen has been known to significantly enhance their chemical reactivity,3,4

although the exact mechanism is not well understood. While a number of

protocols have been used to prepare nitrogen doped carbons,14,15 the more popular

13

methods call for heat treatment of pre-synthesized carbons in an atmosphere of

NH3 or HCN16 followed by exposure to an Ar atmosphere at higher temperatures.

One of the earlier reports by Wang et al17 details the oxidation of Vulcan carbon

using a strong acid and successive heat treatments under NH3 at 600 C and Ar at

900 C. The resulting carbon was found to exhibit enhanced activity for ORR

compared to the undoped carbon as shown by a positive shift of ~200 mV in the

oxygen reduction peak potential.

The enhancement in activity for nitrogenated graphitic carbons has been

attributed to the changes in the electronic structure of the graphene. The

Gooding18,19 and Compton groups20,21 have attributed the enhanced

electrocatalytic activity to the inducement of surface structural disorder

characterized by an increase in density of edge plane sites. While it is well known

that nitrogen doping on carbons produces pyridinic, pyrrolic and quaternary

nitrogen functionalities,11 reports differ on which of these surface functionalities

is responsible for or contributes to ORR activity. Thrower et al22 and Boehm et

al13 have ascribed pyridinic nitrogen functionalities on the edge plane sites to

increased reactivity while other hypotheses23 have proposed nitrogen substituted

in the basal graphene as a vital factor. Literature data have also suggested that the

residual iron from precursors used to synthesize these materials bound to nitrogen

and carbon in a FeNxCy type active site24 is responsible for ORR activity.

Residual iron particles in these carbons have also been attributed to reduction of

peroxide which is one of the byproducts of ORR.25 Although these hypotheses

attempt to elucidate the increased activity of nitrogen doped carbons, none of

14

them satisfactorily explain the exact mechanism of ORR at the active centers of

these carbons.

The ambiguity in assigning how a factor or a combination of factors

contribute to the electrocatalytic activity at nitrogen doped graphitic carbons

originates from complexities involved in the nitrogenation of pre existing carbons.

To this end, a synthetic protocol to prepare and characterize a series of

systematically doped carbon nanotubes was undertaken by Maldonado et al.11

This synthetic strategy allowed for tuning the diameter, length, structure and

composition of nitrogen doped carbon nanotubes (NCNTs).10,11 These studies

elucidated the changes in structure and composition of NCNTs brought upon by

systematic nitrogen doping and enabled correlating these changes with the

chemical reactivity and the electrochemical properties on the NCNTs as compared

to regular undoped carbon nanotubes (undoped CNTs). Parallel studies by Lyon et

al26 dealt with dissolution and passivation of residual Fe encased in NCNTs and

have concluded that FeII/III redox activity and successive passivation of the FeII/III

surface oxides do not impact the onset and activity for ORR at these NCNTs.

This chapter follows up on previous experiments conducted in the

Stevenson research group on systematic nitrogen doping of Fe based carbon

nanotubes (NCNTs). Knowledge gained from those series of experiments was

used to setup a new batch of analyses aimed to clarify the controlling aspects for

ORR at NCNTs. A new variety of NCNTs synthesized using a Ni based precursor

was utilized for these measurements. Efforts were also made to elucidate the

importance of pyridinic nitrogen functionalities comprising total surface nitrogen

15

concentrations on ORR activity at Ni NCNTs. Raman spectral data were used to

quantify the degree of structural disorder on the NCNT surface and account for

the involvement of edge plane sites in ORR.

16

2.2 EXPERIMETAL

2.2.1 Fe Catalyzed CT Synthesis

Nitrogen doped and undoped carbon nanotubes were synthesized using a

floating catalyst CVD method described previously.10,11 Briefly, an automated

dual zone furnace system (Carbolite HST 12/35/200/2416CG) integrated with lab

view controlled mass flow controllers (MKS 1179A) and syringe pumps (New

Era Pump Systems NE-1000) was used in a setup as shown in Figure 2.1. NCNTs

were prepared with pyridine (Fisher) and ferrocene (Alfa Aesar, 20mg ml -1)

precursors (xylene and ferrocene were used for undoped CNTs) loaded in an

airtight syringe (Hamilton 81320) mounted on the syringe pump. The precursors

were injected at a rate of 100µL min-1 into zone 1 of the furnace system

maintained at 130C for NCNTs (150C for undoped CNTs) while the pyrolysis

temperature in zone 2 was 800C (700C for the undoped CNTs). The total gas

flow rate was held constant at 575 sccm for NCNTs and undoped CNTs, and a

calculated feed stream (2.5 – 10 %) of anhydrous ammonia (Aldrich, 99.9%) was

introduced to facilitate higher levels of nitrogen doping with Ar (99.997%

Praxair) making up the reminder of the flow stream. Gas flows for the undoped

CNTs constituted 500 sccm Ar and 75 sccm H2 (99.95% Praxair). The CNTs were

grown on the inside wall of a quartz tube (26mm OD, 22mm ID) that acted as the

reaction chamber.

The as synthesized undoped CNTs and NCNTs were carefully collected

from the inner walls of the quartz reaction chamber and stored in air tight vials

17

Figure 2.1 Schematic of the apparatus used to synthesize undoped and nitrogen doped CNTs.

prior to successive measurements. Electrochemical measurements were conducted

on a measured mass of the CNTs drop cast27 on a glassy carbon rotating disk

Furnace system

Z1 Z2

Mass flow controllers (Ar, H2, NH3)

Syringe pump (Precursors)

smix)

18

electrode (RDE) as detailed in the electrochemical section below. Initial

electrochemical measurements were also performed on CNTs grown directly on a

nickel mesh substrate9 that was cut to specific dimensions so as to keep the

surface area of the electrode constant. The nickel mesh substrates were supported

on a piece of nickel foam and were placed in the center of Z2 of the furnace

system for CNT growth. After the synthesis, these substrates were stored in air

tight vials prior to electrochemical analysis.

2.2.2 i Catalyzed CT Synthesis

Ni catalyzed CNTs were synthesized using the same experimental setup

used for Fe catalyzed CNTs. The precursor mix used for the synthesis consisted of

Nickelocene (Alfa Aesar, 15 mg ml -1) in Acetonitrile (Fisher). The precursor mix

was sonicated for 5 min prior to loading in the injection syringe to facilitate

uniform dispersion of the catalyst in acetonitrile. The resulting mixture was dark

green in color and care was taken to reduce the presence of precipitates and

atmospheric gases in the injection volume. The reaction conditions were similar to

that used for the synthesis of NCNTs.

2.2.3 Electron Microscopy

Transmission electron microscope (TEM) analysis of the CNTs was

performed on a JEOL 2010F instrument operating at 200 kV. The CNT sample

was suspended in anhydrous ethanol and drop cast on a Cu TEM grid covered

with a 3 nm thick amorphous carbon film.

Scanning electron microscope (SEM) analysis was performed using either

a LEO 1530 or a Hitachi S 4000 instrument operating at 10 kV. Carbon samples

19

collected from the interior of the quartz tube were deposited on an adhesive

carbon film prior to analysis. CNTs grown on Ni substrates were introduced per

se into the sample chamber.

2.2.4 BET Surface Area and Pore Size Analysis

Surface area and pore size analysis were performed on the undoped CNTs

and NCNTs using a Quantachrome Autosorb-1 instrument. Sample masses of

atleast 20 mg were introduced into a quartz sample holder and were degassed at

200C for atleast 5 hrs under vacuum. Nitrogen was used as the probe gas.

Surface areas were calculated from an 11 point BET analysis. Micropore

distributions were obtained from density functional theory method and Monte

Carlo simulation methods available within the instrument software.

2.2.5 Raman Spectroscopy

A Renishaw inVia system equipped with a 514.5 nm Ar laser at 3 mW/

cm2 and 100X aperture was used. Spectra were scan averaged for a total time of

300 s. Bands at 1220, 1351, 1487, 1583 and 1624 cm-1 corresponding to I, D, D′′,

G, and D′ bands denoted by Cuesta et al29 were fit using a Peak Fit 4 software

package to correlation factors greater than 0.998. A linear baseline correction was

used to compensate for photoluminescence background.

2.2.6 X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy (XPS) characterization of the samples

was performed using a PHI 5700 ESCA system operating at a scan step size of 0.1

eV and an Al Kα monochromatic line and calibrated with Au 4f7/2, Ag 3d5/2 and

Cu 2p3/2 signals. These spectra were scan averaged 5 times. XPS Spectra were

20

also obtained with a Kratos Axis Ultra DLD system equipped with a dual anode

(150 W Al Kα and Mg sources) with a resolution of 0.1 eV and calibrated with

Au 4f7/2, Ag 3d5/2 and Cu 2p3/2 signals. These spectra were scan averaged 3 times.

Atomic percentages were quantified from survey scans relative to carbon, iron

and nitrogen. FITT 1.2 (Photoelectron Spectroscopy lab, Seoul National

University) software with Shirley background corrections was used to analyze the

spectra.

2.2.7 Thermo Gravimetric Analysis

Thermo gravimetric analysis (TGA) was performed using a Perkin Elmer

7000 TGA. CNT samples (3-5 mg) were held in platinum pans heated to 800 C

in flowing air (Praxair, 99.998%).

2.2.8 Electrochemical Analysis

Electrochemical measurements were carried out on an EG&G Instruments

263A potentiostat equipped with a Pine Instruments MSRX controller for rotating

disk electrode (RDE) measurements. Data acquisition and analysis was performed

on a Corrware (Scribner Associates) software package. Sample slurries prepared

with 0.15 wt. % nafion in anhydrous ethanol and de-ionized water were drop cast

as a film on a glassy carbon (GC) RDE (0.5 cm diameter, Pine Instruments)

polished to a mirror finish and allowed to dry under a stream of Ar (Praxair

99.98%) prior to electrochemical measurements.

Cyclic voltammograms were obtained in a standard three electrode cell

with a gold counter electrode and an Hg/Hg2SO4 reference electrode in O2

(Praxair) saturated 0.1 M H2SO4 prepared with deionized water (>18 MΩ cm). All

21

potentials were converted to NHE for comparison to literature values. Prior to

oxygen reduction, the CNT film was cycled between 0.8 and -0.2V in order to

make the film hydrophilic and achieve steady state voltammograms.

22

2.3 RESULTS AD DISCUSSIO

2.3.1 CT Growth on Substrates

Carbons used throughout this study were undoped and nitrogen-doped

multi walled carbon nanotubes and hereafter referred to as undoped CNTs and

NCNTs respectively. Carbon nanotubes in general are referred to as CNTs. While

a few hypotheses exist as to the mechanism of the growth of CNTs aided by a

metal catalyst, the base catalyzed growth mechanism is commonly accepted. The

graphene layers that make up the CNTs precipitate out from the metal catalyst

after super saturation of the catalyst with gaseous carbon. The graphene layers

either originate out from the metal catalyst, leaving the metal particle at the base

or push the metal catalyst particle out as the growth progresses. In either of these

scenarios the metal particles are encapsulated within the graphene layers.

A variety of electrode supports were chosen for CNT growth to facilitate

the ease and reproducibility of measurements conducted. Initial electrochemical

measurements were carried out on CNTs grown directly on the surface of nickel

mesh electrodes cut to exact dimensions so as to maintain a constant surface area.

SEM characterization of undoped CNTs grown on nickel mesh substrates showed

a remarkable degree of conformal growth and alignment along a perpendicular

axis to the surface of the substrate. Earlier experiments by Maldonado et al.

indicate that the as grown CNTs retain their alignment even after the removal of

the growth substrate by acid dissolution as shown in Figure 2.2.

A majority of electrochemical experiments were carried out on CNTs

drop cast on GC substrates. CNTs used in these experiments were carefully

23

2µm2µm

Figure 2.2 Representative SEM image of CNTs grown directly on a Ni mesh current collector. Image shows that CNTs retain their alignment even after acid dissolution of the Ni.

24

extracted form the inner walls of the quartz tube that was used as a reaction

chamber for CNT growth. The extracted CNTs were in the form of carbon mats.

These carbon mats had a long range order in the case of the undoped CNTs while

they were more powder like in the case of the NCNTs.

2.3.2 TEM Observations of the Fe Catalyst and CT Crystallinity

Ferrocene was used as an Fe precursor that was used as the growth

catalyst for CNTs. TEM observations were carried out on the Fe particles

encapsulated within the undoped CNTs and NCNTs to discern the properties of

the growth catalyst that remained after CNT synthesis. These observations

indicate that the Fe particles were crystalline and were fully encapsulated inside

the graphene matrix in both the undoped CNTs and NCNTs. Figure 2.3 shows

representative TEM images of Fe particles in undoped CNTs and NCNTs. In the

case of the undoped CNTs, the Fe particles are encapsulated within the innermost

walls of the CNTs while the NCNTs appear to grow in either direction of the Fe

catalyst.

The synthesized CNTs were crystalline and found to have diameters

ranging from 20 – 40 nm and lengths of 15 - 20 µm. Undoped CNTs had ordered

sidewalls while NCNTs exhibited disruptions in the graphene stacking consistent

with the incorporation of nitrogen. Nitrogen doping introduces pentagonal defects

causing dislocations in the hexagonal arrangement of the carbon atoms. Nitrogen

doping also caused a noticeable change in the length of crystallinity (Lc) of the

NCNTs, consistent with earlier reports.10

25

Figure 2.3 Representative TEM images of residual Fe catalyst particles encased within the graphitic sheets of CNTs. a) Encapsulated Fe in undoped CNTs. b) Fe encapsulated within 5 % NCNTs.

a) b)

26

2.3.3 Surface Area and Porosity of CTs

Surface area and porosity measurements were conducted on undoped

CNTs and NCNTs to evaluate the effect of an increased surface area and the

presence of micropores on ORR activity. Micropores are commonly defined as

pores that have widths < 25 Å. Mesopores are pores between 25 – 100 Å while

macropores have widths > 100 Å. Porosity and surface area measurements were

carried out on the undoped CNTs and NCNTs using nitrogen as the probe

molecule. These studies were conducted to determine if there was a difference in

the surface area hosted at micropores on NCNTs with increasing nitrogen content

that would be consistent with the findings of the Cai group. Earlier reports by Cai

et al28 discuss the activity of nitrogen doped carbons in relation to the

incorporation of micropores into the carbon matrix. Cai et al conducted a series of

experiments where carbon black was mixed with iron acetate and subjected to

heat treatment at 900 0C under a stream of ammonia. They expected the heat

treatment step to incorporate nitrogen atoms into the carbon and also facilitate the

formation and activation of micropores in the carbon due to the reaction between

the carbon and ammonia. Porosity experiments were also conducted on these

carbons before and after the heat treatment steps and the increased activity of the

resulting carbons was attributed partially to the formation of micropores in the

carbon. Carbons resulting from the heat treatment were found to exhibit no pores

between 8.5 – 9 Å and 21.5 – 22.5 Å while the surface area hosted by pores

between 7 – 22 Å was found to increase within the first 20 min of heat treatment

and subsequently decreased upon further exposure to the heat treatment

27

procedure. A majority ( ~60 %) of the surface area of these carbons was found to

be confined in the micropores while the rest was hosted in mesopores. There was

no evidence of the surface area hosted in macropores.

Figure 2.4 shows a plot of the differential surface area vs. the pore width

for undoped CNTs and NCNTs. It is seen that a majority of the pores in NCNTs

are confined to micropores. Remarkably, micropores at these NCNTs with

varying surface concentrations of nitrogen seem to be confined between 5 – 9 Å.

In the case of the undoped CNTs, the micropores seem to be confined in a

narrower range between 6 – 8 Å. Figure 2.4 also indicates minor differences

between the NCNT varieties at pore sizes between 8 – 20 Å. The presence of a

small Gaussian feature between 12 – 18 Å for the NCNTs indicates that some of

the surface area is confined to pores in that size range. Undoped CNTs were not

found to possess pore sizes greater than 9 Å. There was no evidence of the

presence mesopores or macropores on the CNTs.

BET surface area measurements were also conducted on the undoped

CNTs and NCNTs in order to evaluate the influence of an increase in surface area

on the ORR activity. The undoped CNTs had a surface area of 125 m2 / g while

the 4 % NCNTs had a surface area of 130 m2 / g. Surface area of the NCNTs were

found to increase corresponding to increasing nitrogen content with the 7.5 %

NCNTs having a surface area of 226 m2 / g.

28

2 4 6 8 10 12 140

20

40

60

80

100

120

Differential surface area / m

2Å

−1

g-1

Pore radius / Å

4 % NCNT 5 % NCNT 7.5 % NCNT Undoped CNT

Figure 2.4 Plot of differential surface area corresponding to pore sizes at undoped CNTs and NCNTs measured using nitrogen.

29

2.3.4 Raman Analysis

2.3.4.1 Raman Analysis at Fe Catalyzed CTs

Raman analysis was conducted on Fe catalyzed NCNTs to quantify the

degree of structural disorder on the NCNTs due to the incorporation of nitrogen

on the surface. Previous reports indicate distinct differences in the first order

Raman spectra between the undoped CNTs and systematically doped NCNTs.11

The quantification of D and G bands at 1355 cm-1 and 1585 cm-1 respectively

facilitates study of structural disorder at NCNTs as described by Cuesta et al.29

Ratios of the integrated intensities of the D and G bands were used to calculate

the in plane crystalline length (La) at NCNTs with varying nitrogen doping levels.

Raman analysis also enabled correlations between the degree of structural

disorder and ORR activity at Fe catalyzed NCNTs.10

2.3.4.2 Raman Analysis at i Catalyzed CTs

Raman spectra were acquired for Ni based NCNTs to elucidate the extent

of structural disorder at these carbons in an effort to compare them to Fe based

NCNTs. Figure 2.5 shows a detail of the D and G bands obtained from the Raman

spectra between 1000 - 2000 cm-1 at Ni based NCNTs. A qualitative comparison

of these spectra to Fe based NCNTs showed no discernible differences in the

general features of the D and G bands. Spectra were background subtracted and

the intensities of the D and G band were integrated to calculate ID / IG. This ratio

is a quantitative indicator of the degree of structural disorder at these NCNTs and

can be used to estimate La at these carbons as discussed in earlier sections.30 Edge

plane sites formed at the NCNT surface as a result of the structural disorder have

30

1000 1200 1400 1600 1800 2000

1.20 at.% N

2.96 at.% N

5.25 at.% N

3.73 at.% N

Wavenumbers / cm-1

Figure 2.5 Raman spectra showing the D and G bands on nitrogen doped Ni catalyzed NCNTs.

31

been attributed to an increased activity for ORR. Figure 2.6 shows a plot of the ID

/ IG ratio vs. the pyridinic concentration at Ni based NCNTs. Although there is a

decrease in the length of crystallinity from 3.4 nm to 2.8 nm demonstrated by an

increase in ID / IG for NCNTs with increased nitrogen doping levels, they are

quantitatively smaller than that of the Fe based NCNTs (3.4 nm to 1.7 nm) as

described by previous literature reports.11 These quantitative comparisons were

made for Ni and Fe based NCNTs that had NH3 constituting upto 7.5 % of the gas

flow stream.

2.3.5 XPS Analysis

2.3.5.1 XPS Analysis at Fe Catalyzed CTs

XPS studies on nitrogen doping at Fe catalyzed NCNTs indicate the

existence of three distinct peaks in the nitrogen area at 398.6 eV, 400.9 eV and

404.2 eV. These peaks are attributed to the pyrrolic, pyridinic and quaternary

nitrogen functionalities31 respectively. A fair degree of ambiguity exists in

assigning the peak at 404.2 eV to quaternary nitrogen. Peaks in the near vicinity

have been attributed to physisorbed nitrogen32 in the graphitic lattice and to

chemisorbed nitrogen oxide.33 A systematic study by Maldonado et al. on

increasing the nitrogen content on NCNTs revealed an increase in the pyridinic

fraction of the total nitrogen content on the NCNT surface.11

2.3.5.2 XPS Analysis at i Catalyzed CTs

XPS measurements were performed on systematically doped Ni based

NCNTs in an effort to elucidate the nature of the nitrogen species incorporated on

the surface and enable correlations to ORR activity. The XPS data on four

32

0.0 0.5 1.0 1.5 2.0 2.5 3.0

1.3

1.4

1.5

1.6

1.7

1.8

I D / I G

Pyridinic Nitrogen / at. %

ID / I

G

La

2.4

2.6

2.8

3.0

3.2

3.4

La / n

m

Figure 2.6 Plot of ID / IG vs. the pyridinic concentration at Ni based NCNTs. Also shown is the length of crystallinity (La) correlating degree of structural disorder to the pyridinic fraction of surface nitrogen.

33

different Ni based NCNTs with increasing nitrogen contents showed no

discernable differences in the oxygen and carbon regions when compared to Fe

based NCNTs. The nitrogen region, however, was found to be remarkably

different from that of the Fe based NCNTs.

Figure 2.7 shows XPS spectra with high resolution detail of the nitrogen

region for Ni based NCNTs with incremental levels of total surface nitrogen

concentration. It can be seen that the intensities of peaks assigned to the pyrrolic,

pyridinic and quaternary fractions of surface nitrogen are qualitatively different

from that reported by Maldonado et al11 for Fe based NCNTs. The peak near 404

eV as discussed in preceding sections is hereto referred to as the quaternary peak

although ambiguity exists as to its particular assignment. While the pyrrolic and

quaternary fraction increase almost linearly with an increase in the total nitrogen

concentration, the increase in the intensity of the pyridinic peak is less gradual

across the entire variety of Ni based NCNTs. These observations are in contrast to

observations at Fe based NCNTs as detailed in the subsequent discussion.

Constant flow rates of NH3 used in Ni catalyzed and Fe catalyzed NCNT

synthesis for each variety of carbon resulted in a lower level of nitrogen doping at

Ni catalyzed NCNTs. For example when NH3 constituted 7.5 % of the gas flow

stream during Fe catalyzed NCNT synthesis, the resulting carbon had 7.5 at. %

nitrogen incorporated on the surface. Under the same conditions for Ni catalyzed

NCNT synthesis, the resulting carbon only had 3.7 at. % nitrogen incorporated on

the surface.

34

390 395 400 405 410 415

1.20 at.% N

2.96 at.% N

3.73 at.% N

5.25 at.% N

Binding Energy / eV

Figure 2.7 XPS spectra showing the nitrogen region in nitrogen doped Ni catalyzed NCNTs.

35

Quantitative measurements were made on the XPS spectra obtained on the

Ni based NCNTs by integrating individual peaks that contributed to the total

nitrogen concentration against a Shirley background correction. Literature reports

and earlier experiments at the Stevenson research lab have strongly suggested the

surface concentration of pyridinic nitrogen as a major factor in promoting ORR at

Fe catalyzed NCNTs. A comparison of Ni based NCNTs and Fe based NCNTs

that had comparable total surface nitrogen content (4 % NCNT – Fe vs. 3.7 %

NCNT – Ni), showed that the distribution of individual nitrogen species

(pyridinic, pyrrolic and quaternary) that made up the total concentration was

found to be significantly different. Quaternary nitrogen concentrations were found

to be relatively minor at 1.2 % NCNTs and 3 % NCNTs but increase dramatically

at 3.7 % NCNTs and 5.3 % NCNTs. Earlier reports11 see a rise in the intensity of

the pyrrolic peak at higher overall nitrogen concentrations, but there are no

reports of a dramatic rise in the concentration of quaternary/ physisorbed nitrogen.

Figure 2.8 indicates that as the surface nitrogen content increases, the

pyrrolic and quaternary fractions in Ni based NCNTs rise in intensity and

constitute a majority of the nitrogen content. In the case of Fe based NCNTs, the

pyridinic fraction makes up more than 50 % of the total surface nitrogen

concentration of NCNTs with total nitrogen concentrations upto 7.5 at. % (3.3 at.

% pyridinic nitrogen in Fe catalyzed 7.5 % NCNTs). In contrast to the

observations at Fe based NCNTs, Ni based NCNTs seem to support pyridinic

fractions that constitute an average of 40 % of the total nitrogen content across a

variety of the NCNTs doped with systematically increasing concentrations of

36

0 1 2 3 4 5 60.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0 Pyridinic nitrogen Pyrrolic and Quaternary nitrogen

Total Nitrogen / %

Pyridinic nitrogen / at. %

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Pyrro

lic and Quaternary n

itrogen / a

t. %

Figure 2.8 Quantitative comparison of the different nitrogen fractions in nitrogen doped Ni catalyzed NCNTs based on XPS spectra.

37

nitrogen as shown in Figure 2.8. The relative concentration of the pyridinic

fraction decreases with increasing nitrogen doping levels with the pyridinic

fraction making up 2.1 at. % of the total nitrogen concentration at 5.2 % NCNTs.

These observations clearly indicate the differences between nitrogen doping at Fe

based NCNTs and Ni based NCNTs especially with regard to the pyridinic

nitrogen content.

2.3.6 Effect of Increasing Fe content in CTs on ORR

This section details experiments performed to discern effects of the Fe

catalyst in CNTs on ORR. The first set of experiments was carried out to study

the effect of Fe bound to the surface of the CNTs on ORR. A pre-determined

mass of 4 % NCNTs was drop cast onto a GC electrode and control ORR studies

were conducted in oxygen saturated 0.1 M H2SO4. The electrolyte solution was

then infused with predetermined volumes of iron sol prepared in accordance with

Sorum et al.34 Fe2+ spikes amounting to 1, 2 and 5 mM were injected into the

electrolyte with the ORR being run after every subsequent spike. The resulting

ORR peak potentials were then compared to the control experiment performed on

the blank carbon.

Cyclic voltammograms of ORR performed on the 4 % NCNT with

varying Fe spikes is shown in Figure 2.9. It was seen that while there was a slight

increase in the current density for ORR in the presence of an Fe spike, there was

no effective shift in the peak potential. The increase in current densities were also

found to plateau after the first 1 mM spike suggesting no strong correlations to the

38

1.2 1.0 0.8 0.6 0.4 0.2 0.0

0.00

-0.03

-0.06

-0.09

-0.12

-0.15

O2 sat. 0.1M H

2SO

4

ν = 20mV / s

Current / mA

Potential / V vs. NHE

4 % NCNT 1mM spike 4 % NCNT 2mM spike 4 % NCNT 5mM spike Control 4 % NCNT

Figure 2.9 Oxygen reduction at 4 % NCNTs with varying concentrations of Fe2+ added to the electrolyte solution.

39

presumed effect on an increased concentration of surface bound Fe2+ to an

enhanced activity for ORR.

Experiments were also performed to deduce whether an increased

concentration of Fe in the precursor mix led to an enhanced activity for ORR at 4

% NCNTs. Accordingly, the concentration of Fe in the precursor mix was

increased in predetermined amounts resulting in mass loadings of Fe in the as

prepared NCNTs from 11 wt. % to 22.6 wt. %. Fe mass loading in the 4 %

NCNTs was measured using TGA analysis. The resulting carbons were analyzed

for their activity towards ORR. It was seen that there was a negative effect to an

increase in the mass of Fe in NCNTs. Figure 2.10 shows a plot of the peak

potential for ORR for each of the 4 % NCNTs synthesized using varying masses

of Fe in the precursor mix. The peak potential for ORR shifts negative by ~ 100

mV with an increase in 10 wt. % Fe. XPS measurements conducted on the same

carbon samples indicated that there was a decrease in the pyridinic nitrogen

fraction by 0.5 at. % corresponding to a decrease in total surface nitrogen content

by 1.2 at. % between NCNTs with Fe loadings of 11 wt. % and 22.6 wt. %. These

observations suggest a strong correlation between oxygen reduction efficiency

and the pyridinic surface nitrogen content rather than Fe content in the NCNTs.

2.3.7 Electrochemical Analysis at CTs

2.3.7.1 ORR at Fe catalyzed CTs

The activity of Fe catalyzed NCNTs and undoped CNTs for ORR was

tested in oxygen saturated 0.1 M H2SO4. Figure 2.11 shows representative cyclic

voltammograms for oxygen reduction at undoped CNTs and 4 % and 7.5 %

40

30 35 40 45 50 55 60

0.30

0.32

0.34

0.36

0.38

0.40

0.42


Fe loading (mg) in NCNF Precursor

Ep in O2 sat. 0.1 M H

2SO

4

Figure 2.10 Plot showing correlation between peak potential for ORR activity corresponding to mass of iron precursor used to synthesize NCNTs.

41

NCNTs. The peak potentials for ORR at 4 % and 7.5 % NCNTs are shifted

positive by 580 mV and 730 mV respectively compared to the undoped CNTs.

Although the commonly accepted explanation for this increase in activity in

NCNTs is attributed in general to nitrogen doping, quite a few hypotheses exist as

to the exact nature of the active sites that adsorb and reduce oxygen.

Attempts to find a correlation between the surface area and ORR activity

at the NCNTs and undoped CNTs were not successful, contradictory to previous

reports that attempted to explain enhanced activity at nitrogenated carbons to

increased surface areas.28 While the difference in surface areas between the

undoped CNTs and 4 % NCNTs was only 5 m2 / g, the peak potential for ORR in

0.1 M H2SO4 was shifted positive by 580 mV. Peak potentials were shifted

positive by 150 mV between 4 % NCNTs and 7.5 % NCNTs although the

difference in surface areas was 96 m2 / g indicating no strong correlations

between surface area and activity at undoped CNTs and NCNTs. While the

undoped CNTs and NCNTs had pore size distributions within the same range as

described in a previous section dealing with surface area and porosity, the ORR

activity at NCNTs is significantly higher than undoped CNTs as shown in Figure

2.11. These observations suggest that the presence of micropores at these carbons

does not seem to be an influencing factor in the enhanced activity for ORR at

NCNTs.

The cyclic voltammetry data viewed in conjunction with the Raman

analysis reveals that there was an inverse relationship between the activity for

ORR at CNTs and NCNTs with La values at these carbons. This is evidenced by a

42

0.6 0.4 0.2 0.0 -0.2 -0.4-0.1

0.0

0.1

0.2

Current / mA

Potential /V vs. NHE

Undoped CNT 4 % NCNT - Fe 7.5 % NCNT - Fe

O2 sat. 0.1 M H

2SO

4

ν = 20 mV / s

Figure 2.11 ORR activity for undoped CNTs and NCNTs in oxygen saturated 0.1 M H2SO4. ν = 20 mVs-1

43

decrease in La from 3.4 nm to 1.7 nm for the 4 % NCNTs and 7.5 % NCNTs

respectively. These studies provide strong arguments that systematic induction of

structural disorder on the NCNT surface characterized by an increase in edge

plane sites corresponding to nitrogen doping plays an important role in the

adsorption and reduction of oxygen at these sites.

2.3.7.2 ORR at i Catalyzed CTs

Figure 2.12 shows representative cyclic voltammograms for ORR at Ni

catalyzed NCNTs containing 3.7 at. % N compared to Fe based undoped CNTs

and NCNTs containing 4 and 7.5 at. % N in oxygen saturated 0.1 M H2SO4. The

Ni catalyzed NCNTs are more active for ORR compared to Fe catalyzed undoped

CNTs. The peak potential for ORR was shifted positive by ~360 mV for the Ni

catalyzed NCNTs against the undoped CNTs. The peak potential for ORR at Fe

based 4 % NCNTs was shifted positive by ~ 220 mV compared to the Ni based

NCNTs. While the 3.7 % Ni based NCNTs had a slightly lower overall surface

nitrogen concentration than the 4 % NCNTs, the pyridinic fraction at the Ni

NCNTs was found to be 1.5 at. % compared to 2.1 at. % found in Fe based 4 %

NCNTs offering a correlation for ORR activity to the amount of pyridinic

nitrogen.

The voltammetry data correlate well with Raman observations at the Ni

based NCNTs. Raman analysis suggests a lower degree of structural disorder

corresponding to an increase in surface nitrogen content at Ni based NCNTs

compared to Fe based NCNTs (as detailed in the Raman analysis section). This

results in a smaller number of edge plane sites that are more active than regular

44

0.8 0.6 0.4 0.2 0.0 -0.2 -0.4-0.05

0.00

0.05

0.10

0.15

0.20

0.25

Current / mA


Undoped CNT 3.7 % NCNT - Ni 4.0 % NCNT - Fe 7.5 % NCNT - Fe

O2 sat. 0.1 M H

2SO

4

ν = 20 mV / s

Figure 2.12 Comparison of cyclic voltammograms for ORR at undoped CNTs and NCNTs synthesized using Fe and Ni precursors in oxygen saturated 0.1 M H2SO4. ν = 20 mVs-1.

45

basal plane sites for adsorption and reduction of oxygen. Figure 2.13 shows a plot

of 1 / La values calculated for Fe and Ni based NCNTs with respect to Ep values

for ORR at the corresponding carbons. A direct correlation between these two

parameters emphasizes the effect of edge plane sites on ORR activity.

2.3.7.3 Effect of Pyridinic itrogen on ORR

A plot correlating the amount of pyridinic nitrogen at Ni and Fe based

NCNTs to the peak potential for ORR at these carbons is shown in Figure 2.14.

The pyridinic fraction was calculated for each of the NCNTs by performing a

quantitative analysis of the peak near 400 eV with respect to the total nitrogen

concentration on the NCNT surface. It is seen that the peak potential for ORR

shifts by ~750 mV corresponding to an increase of 2.7 at. % pyridinic nitrogen

across the entire variety of carbons. This corresponds to a positive shift of ~ 295

mV with respect to an increase in each at. % of pyridinic nitrogen in the Fe based

NCNTs. No strong correlations were found between pyrrolic and quaternary

nitrogen contents with ORR activity.

Figure 2.14 shows that the Ni based NCNTs and the Fe based NCNTs

have ORR activities that track well with the pyridinic nitrogen content (R2 =

0.99). A lesser degree of correlation (R2 = 0.94) was seen with ORR activities

compared to total nitrogen content as seen in Figure 2.15. These results are a clear

indication that the ORR activity at these NCNTs is independent of the Fe or Ni

catalyst that was used in the synthesis of these NCNTs. Earlier reports by

Maldonado et al.11 have correlated the amount of pyridinic fraction of the total

nitrogen content to the amount of edge plane sites created due to the nitrogen

46

0.1 0.2 0.3 0.4 0.5 0.6 0.7

-0.2

0.0

0.2

0.4

0.6

1 / La

Ep / V vs. NHE

1 / La (nm

-1)

R2 = 0.99

Figure 2.13 Plot showing correlation between peak potential for ORR activity corresponding to edge plane content calculated from Raman spectra at Fe and Ni based NCNTs. ORR activity was measured in oxygen saturated 0.1 M H2SO4.

47

0.0 0.5 1.0 1.5 2.0 2.5 3.0

-0.2

0.0

0.2

0.4

0.6

R2 = 0.99

Ep /V vs. NHE

Pyridinic Nitrogen / at. %

Undoped CNT - Fe

1.2 % NCNT - Ni

2.9 % NCNT - Ni

3.7 % NCNT - Ni

4 % NCNT - Fe

Pyridinic nitrogen

Slope = 295 mV / at. % 7.5 % NCNT - Fe

Figure 2.14 Plot showing correlation between peak potential for ORR activity corresponding to the pyridinic nitrogen fraction at Fe and Ni based NCNTs. ORR activity was measured in oxygen saturated 0.1 M H2SO4.

48

0 2 4 6 8

-0.2

0.0

0.2

0.4

0.6

Undoped CNT - Fe1.2 % NCNT - Ni

4 % NCNT - Fe

7.5 % NCNT - Fe

Ep /V vs. NHE

Total Nitrogen / at. %

Total nitrogen

R2 = 0.94

2.9 % NCNT - Ni

3.7 % NCNT - Ni

Figure 2.15 Plot showing correlation between peak potential for ORR activity corresponding to the total nitrogen content at Fe and Ni based NCNTs. ORR activity was measured in oxygen saturated 0.1 M H2SO4.

49

doping in conjunction to Raman analysis at NCNTs. These background studies

and the present set of data as shown in Table 2.1 compile a strong correlation for

ORR activity to the pyridinic nitrogen content and the presence of edge plane

sites at the NCNT surface.

51

2.4 COCLUSIOS

The ability to tune the physicochemical properties such as length,

diameter, structural orientation and composition of as grown Fe or Ni catalyzed

NCNTs was established. Structural and compositional variations brought upon by

systematic nitrogen doping on CNTs as quantified by Raman and XPS analysis

were correlated to the enhanced activity for ORR at the surface of these NCNTs.

These studies provide fundamental knowledge that can be used in understanding

and manipulating NCNTs for use as metal catalyst supports for ORR as dealt with

in subsequent chapters.

52

2.5 REFERECES

1. Planeix, J. M.; Coustel, N.; Coq, B.; Brotons, V.; Kumbhar, P. S.; Dutartre, R.; Geneste, P.; Bernier, P.; Ajayan, P. M. J. Am. Chem. Soc. 1994, 116, 7935.

2. McCreery, R. L. Carbon Electrodes: Structural Effects on Electron Transfer Kinetics. In Electroanalytical Chemistry; Bard, A. J., Ed.; Dekker: New York, 1991; Vol. 17.

3. Glenis, S.; Nelson, A. J.; Labes, M. M. Journal of Applied Physics 1999, 86, 4464.

4. Endo, M.; Hayashi, T.; Hong, S. H.; Enoki, T.; Dresselhaus, M. S. J. Appl. Phys. 2001, 90, 5670.

5. Berber, S; Kwon, Y-K.; Tomanek, D. Phys. Rev. Lett. 2000, 84, 4613.

6. Nevidomskyy, A. H.; Csanyi, G.; Payne, M. C. Phys. Rev. Lett. 2003, 91, 105502.

7. Kaneko, K.; Ishii, C.; Ruike, M.; Kuwabara, H. Carbon, 1992, 30, 1075.

8. Sharda, T.; Soga, T.; Jimbo, T.; Umeno, M. Diamond Relat. Mater. 2000, 9, 1331.




12. Brezina, M.; Jindra, J.; Mrha, J. Coll. Czech. Chem. Comm. 1967, 33, 2363.

13. Boehm, H. P.; Mair, G.; Stoehr, T.; DeRincon, A. R.; Tereczki, B. Fuel 1984, 63, 1061.

14. Lahaye, J.; Nanse, G.; Bagreev, A.; Strelko, V. Carbon 1999, 37, 585.

15. Sjostrom, H.; Stafstrom, S.; Boman, M.; Sundgren, J. E. Phys. Rev. Lett. 1995, 75, 1336.

16. Stohr, B.; Boehm, H. P.; Schlogl, R. Carbon 1991, 29, 707.

53

17. Wang, H.; Cote, R.; Faubert, G.; Guay, D.; Dodelet, J. P. J. Phys. Chem. B 1999, 103, 2042.

18. Gooding, J. J. Electrochimica Acta 2005, 50, 3049.

19. Chou, A.; Bocking, T.; Singh, N. K.; Gooding, J. J. Chem. Commun. 2005, 7, 842.

20. Banks, C. E.; Moore, R. R.; Davies, T. J.; Compton, R. G. Chem. Commun. 2004, 16, 1804.

21. Salimi, A.; Banks, C. E.; Compton, R. G. Analyst, 2004, 129, 225.

22. Strelko, V. V.; Kuts, V. S.; Thrower, P. A. Carbon 2000, 38, 1499.

23. Sidik, R. A.; Anderson, A. B.; Subramanian, N. P.; Kumaraguru, S. P.; Popov, B. N. J. Phys. Chem. B 2006, 110, 936.

24. Jouen, F.; Charreteur, F.; Dodelet, J. P. J. Electrochem. Soc. 2006, 153, A689.

25. Sljukic, B.; Banks, C. E.; Compton, R. G. -ano Lett. 2006, 6, 1556.

26. Lyon, J. L.; Stevenson, K. J. Langmuir 2007, in press.

27. Schmidt, J.; Gasteiger, H. A; Stab, G. D.; Urban, P. M.; Kolb, D. M.; Behm, R. J. J. Electrochem. Soc. 1998, 145, 2354.

28. Jaouen, F.; Lefevre, M.; Dodelet, J. P.; Cai, M. J. Phys. Chem. B 2006, 110, 5553.

29. Cuesta, A.; Dhamelincourt, P.; Laureyns, J.; Martinezalonso, A.; Tascon, J. M. D. Carbon 1994, 32, 1523.

30. Sjostrom, H.; Stafstrom, S.; Boman, M.; Sundgren, J. E. Phys. Rev. Lett. 1995, 75, 1336.

31. Stohr, B.; Boehm, H. P.; Schlogl, R. Carbon, 1991, 29, 707.

32. Choi, H. C.; Park, J.; Kim, B. J. Phys. Chem. B 2005, 109, 4333.

33. Biniak, S.; Szymanski, G.; Siedlewski, J.; Swiatkowski, A. Carbon 1997, 35, 1799.

54

34. Sorum, C. H. J. Am. Chem. Soc. 1928, 50, 1263.

55

CHAPTER 3

Synergistic Assembly of Dendrimer Templated Catalysts and itrogen Doped Carbon anotube Electrodes for

Oxygen Reduction

3.1 ITRODUCTIO

High surface area carbons such as Ketjenblack and Vulcan carbon used as

supports for electrocatalysis applications are beneficial in terms of providing

electronic conductivity and high dispersion of metal catalysts.1,2 Unfortunately,

the multitude of preparation strategies for carbon supported catalysts makes it

difficult to understand the role of the carbon support on electrocatalysis including

the degree of catalyst utilization, promotion of catalyst-support interactions, and

the stability of the catalyst towards dissolution, agglomeration, and other

degradation processes. Moreover, carbon supports are typically prepared via

aggressive processes for activation including refluxing in concentrated acids3

(HNO3, H2SO4, HCN) or strong oxidizing agents4 (H2O2, KMnO4) to create

surface functionalities (carbonyl, carboxylate, ester-like oxygen, alcohol) to

facilitate more efficient anchoring and loading of the metal catalyst via

impregnation5, coprecipitation

6, microemulsion

7 or sonochemical

8 methods. The

activation methods often significantly degrade the preferred structural and

compositional properties of both the carbon support and active metal catalyst and

have typical drawbacks of large average catalyst size, broad size distribution, and

poor reproducibility.9 Of vital importance is the need to significantly improve

56

catalyst dispersion and utilization (typically <20% is catalytically active) on the

carbon support, and to reduce catalyst sintering and poisoning processes. Carbon

nanotubes (CNTs) have attracted significant interest as catalyst supports after

their recent discovery,10 as they have optimal electronic conductivity, proper

surface area and pore structure.11,12,13,14

While as synthesized CNTs have some of

the desirable properties required for supports, efficient anchoring and utilization

of loaded metal catalysts is still achieved through extensive surface modification

of the CNTs15,16,17

using some of the aggressive and time consuming conditioning

protocols listed above. To this end, we have been exploring facile synthetic

strategies for preparing both robust, high surface area nitrogen doped CNT

supports18,19

and active metal catalysts with well controlled properties that

circumvent the majority of processing steps required by more traditional loading

routes, such as washing, drying, calcination, and reduction.

This chapter describes the use of nitrogen doped carbon nanotube (NCNT)

supports 19 that strongly bind size monodisperse mono- and bimetallic dendrimer

encapsulated catalysts prepared by a dendrimer template method20 without

necessitation for covalent21 or non-convalent

22 functionalization of the carbon

support prior to catalyst loading. The dendrimer template method of synthesizing

nanoparticles is versatile since a variety of nanoparticle properties such as size,

structure and composition23 can be modified with relative ease compared to

various other methods24,25,26

employed to synthesize nanoparticles. The ability to

modify properties of the interior and exterior functional groups in a dendrimer

template facilitates control of the way in which precursor metal ions coordinate

57

within the dendrimer template. This is due to a difference in basicity between the

interior and exterior functional groups in the dendrimer. For example, in the case

of amine terminated dendrimers, the interior amines have a pKa value of 6.30

while the exterior amines have a pKa value of 9.23 which facilitates easier

protonation of the exterior amines and thus allows for selective coordination of

metal ions with the interior amines. The presence of specific functionalities on the

dendrimer exterior also provides an efficient way to anchor the resultant

encapsulated nanoparticle to a specific substrate. For example the Crooks group

has shown that amine terminated PAMAM dendrimers show strong adsorption to

HOPG.27 The use of a particular dendrimer generation allows for efficient control

of size of the nanoparticle based on the number of available interior functional

groups that coordinate with metal ions from the precursor.23,24,25

While the

diameter of the dendrimer molecule increases logarithmically with each

dendrimer generation, the number of functional groups increases exponentially.

For this particular study, PAMAM generation 4, amine terminated dendrimers

(G4-NH2) were used. The number (62) of interior functional groups in G4

dendrimers was best suited to synthesize ~2 nm size nanoparticles and the amine

terminated peripheral functional groups allowed for facile adsorption onto

NCNTs.

Scheme 1 demonstrates the basic approach for the assembly of the CNT-

DEN composites. The advantage of this scheme is that the compositional and

structural properties of both the carbon support 18 and dendrimer encapsulated Pt

58

Scheme 3.1 Preparation of amine terminated PAMAM dendrimer encapsulated Pt nanoparticles (G4-NH2 Pt-DENs) and adsorption onto carbon nanotube (CNT) supports.

59

nanoparticle (Pt-DEN) catalyst can be reproducibly prepared and synergistically

tuned to directly assemble carbon-supported catalyst composites via adsorption

from aqueous solution. This chapter focuses on the uniform dispersion and

loading of dendrimer encapsulated mono- and bimetallic catalysts supported on

undoped CNTs and NCNTs that show enhanced catalytic behavior for the oxygen

reduction reaction (ORR) in fuel cells. The ability to prepare high quality, well

dispersed supported metal catalysts, with sizes in the 2-5 nm range that avoid

painstaking and lengthy synthetic and post processing activation steps is required

to advance electrocatalysis performance.

60

3.2 EXPERIMETAL

3.2.1 Synthesis of Monometallic Dendrimer Encapsulated anoparticles

Fourth generation, amine terminated, poly(amidoamine) PAMAM

dendrimers (G4-NH2) (Dendritech) were used as templates to prepare Pt-DENs

and Pd-DENs. The G4-NH2 templates were received in a 10 % methanol solution

that was removed under vacuum prior to DEN synthesis. K2PtCl4 and K2PdCl4

(Aldrich) were used as precursors for Pt and Pd DENs respectively.27 A carefully

measured volume of G4-NH2 dendrimers was dried off methanol and suspended

in a pH adjusted HCl solution (pH 5 for Pt-DENs and pH 3 for Pd-DENs)

resulting in 5 – 20 µM dendrimer solutions. This pH adjustment facilitated in

selective protonation of the exterior amines. Aqueous 0.1 M K2PtCl4 or 0.1 M

K2PdCl4 constituting 40 mol equivalent of the dendrimers was added to the

dendrimer solution. A slight increase in pH resulting from the addition of the

metal precursors was countered with addition of small aliquots of dilute 0.1 M

HCl until the acidity was restored to the original pH levels. The resulting solution

was left to stir for 72 hrs due to sluggish co-ordination times between the Pt ions

and interior amines.28,29

Pd co-ordination rates were much faster with the reaction

happening within 30 min. A 20 mol excess (w.r.t metal ion concentration in

solution) of 0.5 M NaBH4 (Fisher) was added drop by drop after allowing for

appropriate co-ordination times. The resulting DEN solutions were dark brown in

color which is consistent with the formation of nanoparticles by reduction. After

reduction the DEN solutions were adjusted to pH 8 to prevent agglomeration of

the nanoparticles. Following pH adjustment the DEN solutions were subjected to

61

dialysis against 10 L of 18 MΩ cm nanopure water for 12 hrs to remove

impurities resulting from the synthesis. Cellulose dialysis sacks with MWCO

12000 (Sigma) were employed in the dialysis step.30 Dialyzed DEN solutions

were stored in acid washed airtight vials prior to further analysis.

3.2.2 Synthesis of Bimetallic Dendrimer Encapsulated anoparticles

Bimetallic PdAu DENs were synthesized primarily to compare ORR

activity at a non Pt based catalyst and to gain insight into ORR activity at a Pd

alloy surface. A carefully measured volume of G4-NH2 dendrimers was dried off

methanol and suspended in an aqueous HCl solution that was adjusted to pH 3

resulting in a 20 µM dendrimer solution. A 0.1 M K2PdCl4 solution constituting

20 mol equivalent of the dendrimers was added to the dendrimer solution and the

pH was adjusted back to 3. Pd ions were allowed to coordinate with the interior

amines for 30 min after which a 20 mol equivalent of 0.1 M HAuCl4 was added

and the solution allowed to stir for an additional 10 min. A 20 mol excess 0.5 M

NaBH4 solution was added drop by drop after stirring was completed and pH of

the resultant DEN solution was adjusted to 8 to prevent agglomeration. The

resulting DEN solution was brown in color and was dialyzed against 10 L of 18

MΩ nanopure water for 12 hrs to remove impurities resulting from the synthesis.

The same type of dialysis sacks used for the monometallic DENs (cellulose with

12000 MWCO) was used in the dialysis step. PdAu DEN solutions were stored in

acid washed airtight vials prior to further analysis. While a number of synthesis

protocols31 like co-complexation and sequential loading using quaternized

62

dendrimers have been utilized to prepare bimetallic particles, this present scheme

was suited for studies on ORR activity at PdAu surfaces.

3.2.3 Synthesis of Undoped CTs and CTs

CNTs were prepared using the floating catalyst chemical vapor deposition

method18,19

described in chapter 2. Briefly, xylene (Aldrich) as carbon source and

ferrocene (Aldrich) as catalyst were used for the growth of the undoped CNTs

while a pyridine (Fisher) and ferrocene precursor combination was used for the

synthesis of nitrogen-doped CNTs. The use of a pyridine precursor allowed for a

controlled doping of ~ 4 at. % N on the NCNT surface. Higher surface nitrogen

concentrations (5-10 at. % N) were obtained by introducing a regulated stream of

ammonia gas (Aldrich) into the CVD furnace system along with the pyridine and

ferrocene precursors. As grown undoped CNTs and NCNTs were carefully

extracted from the inner walls of the quartz tube used as a reaction chamber and

stored in airtight vials prior to characterization and further use.

3.2.4 Adsorption of Pt-DEs on CTs and CTs

Pt-DEN standard solutions ranging from 5-20 µM were prepared and

characterized using UV-Vis spectroscopy using a Varian Cary 5000 instrument by

monitoring the featureless absorption at 350 nm that reports on the formation and

concentration of colloidal Pt. A calibration curve from 0-20 µM was constructed

to follow the adsorption of the Pt-DENs onto the various CNT supports by

monitoring the decrease in absorbance of the supernatant at 350 nm as a function

of immersion time from 1 to 24 h for a 1 mg/ml carbon sample dispersed in a

stirred solution containing known amounts of Pt-DENs.32 The absorbance at 350

63

nm has been attributed to the presence of colloidal Pt resulting from the

synthesis.27 A calibration curve was constructed from absorbance values at 350

nm from Pt-DEN solutions of varying concentrations from 5-20 µM.

Absorbance readings were taken every 30 min during the adsorption

process by first centrifuging the solution for 2 min so as not to withdraw any of

the CNTs suspended in solution. A 3 ml volume of the supernatant was then

carefully withdrawn from solution and placed in a quartz cuvette used for

absorbance measurements. After the measurements were completed, the

supernatant was injected back into the CNT/ DEN solution and the adsorption

process was allowed to continue accompanied by stirring. Nylon filters (GE corp.)

with an average of 0.2 µm pore size were employed in the filtration process.

Background experiments revealed that the filters did not adsorb any significant

quantities of the DENs. After the initial 3 hrs, measurements were taken every

hour since the adsorption slowed down considerably. While the NCNTs used in

this experiment were hydrophilic and went into solution immediately after

introduction, the undoped CNTs were not and took longer times to be completely

wet in solution. Care was taken to ensure that CNTs were well dispersed before

they were introduced into the DEN solution. In the case of 7.5 % NCNTs, the

DEN solution was regenerated with a fresh solution after absorbance

measurements indicated complete adsorption of DENs from the initial solution.

These studies were conducted to measure the maximum amount of adsorption of

DENs at the NCNTs. Once the adsorption process was deemed complete, the

CNT / DEN solutions were centrifuged and the supernatant was extracted

64

carefully. The resultant CNT/ Pt-DEN composites were washed with de ionized

water and allowed to dry. Pt-DEN composites were stored in air tight vials prior

to further analyses.


Transmission electron microscope (TEM) analysis of the CNTs was

performed on a JEOL 2010F instrument operating at 200 kV. DEN samples were

drop cast on a Cu TEM grid covered with a thin layer of amorphous carbon and

allowed to dry overnight in a dessicator prior to TEM analysis. CNT samples with

adsorbed DENs were allowed to dry after recovery from the DEN solution and

suspended in anhydrous ethanol prior to drop casting on a lacey carbon covered

Cu TEM grid.



7000 TGA. CNT/ DEN samples (3-5 mg) were held in platinum pans heated to

800 C in flowing air (Praxair, 99.998%) at a rate of 5

0C / min. The resulting

mass contained Fe2O3 from the residual Fe catalyst in the CNTs and Pt from Pt-

DENs. Control TGA studies performed using particular varieties of the blank

CNTs enabled the mass of hematite to be subtracted facilitating calculations of the

mass loading of Pt on CNTs.

3.2.7 Electrochemistry at CT/ DE Composites

Electrochemical measurements were carried out on an EG&G Instruments



65

on a Corrware (Scribner Associates) software package. CNT/ DEN sample

slurries prepared with 0.15 wt. % nafion in anhydrous ethanol and de-ionized

water were drop cast as a film on a glassy carbon (GC) RDE (0.5 cm diameter,

Pine Instruments) polished to a mirror finish and allowed to dry under a stream of

Ar (Praxair 99.98%) prior to electrochemical measurements.

Cyclic voltammograms were obtained in a standard three electrode cell

with a gold counter electrode and an Hg/Hg2SO4 reference electrode in O2

(Praxair) saturated 0.1 M H2SO4 prepared with deionized water (>18 MΩ cm). All

potentials were converted to NHE for comparison to literature values. Prior to

oxygen reduction, the CNT/ DEN film was cycled between 0.8 and -0.2V in order

to make the film hydrophilic and achieve steady state voltammograms. This step

was primarily used for undoped CNT/ DEN catalyst films that required atleast 10

CV cycles before exhibiting hydrophilicity. NCNTs on the other hand are easily

wettable due to the presence of positively charged nitrogen groups on the surface

and to the existence of more edge plane sites than the undoped CNTs.

Rotating disk electrode measurements were performed on the same films

after the completion of cyclic voltammetry measurements. The electrode was

rotated between 250 and 2000 rpm in increments of 250 rpm except for the

potentiodynamic voltammogram obtained at 1600 rpm.33 The electrolyte solution

was replenished with O2 in between two potentiodynamic measurements and also

between consecutive CVs. Potentials were typically scanned from 1 V vs. NHE to

0 V vs. NHE.

66


3.3.1 Structure and Composition of DEs

A representative TEM image in Figure 3.1 shows G4 NH2 Pt-DENs from a

20 µM solution. The particles are well dispersed and do not exhibit any

agglomeration. The corresponding histogram reveals that the particle size

distribution is narrow and nearly monodisperse consistent with earlier reports.27

The average diameter of the Pt nanoparticles was found to be 2.2 ± 0.3 nm. Pd-

DENs had an average diameter of 2.0 ± 0.4 nm (Figure 3.2) while the PdAu

DENs (Figure 3.3) had average diameters of 2.3 ± 0.4 nm. Energy dispersive

spectroscopy (EDS) analysis performed on single PdAu nanoparticles (Figure 3.4)

showed that these were bimetallic and had compositions of Pd 31 ± 5 % and Au

69 ± 5 %. The standard deviations result from challenges associated with

obtaining reliable EDS spectra from particles smaller than 2 nm.31 The TEM

images also show that the DENs are crystalline as evidenced from observations of

lattice structures.

3.3.2 Analysis of Pt-DE Adsorption at Undoped CTs and CTs

Pt-DEN adsorption measurements were performed on undoped CNTs and

NCNTs by monitoring the absorbance at 350 nm over 24h. Figure 3.5 shows a

representative adsorption profile for Pt-DENs on 4 % NCNTs over a 24h period.

The absorbance between 200- 800 nm is seen to decrease consistently as a

function of time corresponding to a lowering of the DEN concentration in the

supernatant. The broad absorbance between 300 – 800 nm has been associated

with the presence of colloidal Pt and is seen to decrease until no significant

67

Figure 3.1 Representative TEM image of Pt –DENs. The histogram depicts a typical particle size distribution for the Pt-DENs.

1.4 1.6 1.8 2.0 2.2 2.40

5

10

15

20

25

30

Co

un

ts

Particle Diameter /nm

68

Figure 3.2 Representative TEM image of Pd –DENs. The histogram depicts a typical particle size distribution for the Pd-DENs.

1.4 1.6 1.8 2.0 2.2 2.40

5

10

15

20

25

30

Co

unts

Particle Diameter / nm

69

Figure 3.3 Representative TEM image of PdAu –DENs. The histogram depicts a typical particle size distribution for the PdAu-DENs.

1.4 1.6 1.8 2.0 2.2 2.4 2.60

5

10

15

20

25

30

Cou

nts

Particle Diameter / nm

70

Figure 3.4 Representative energy dispersive spectrum for PdAu- DENs.

Au

Au PdPd Au

Au

Pd

Au

Au

Au

Au

Pd Au

Au

Pd

Au

Au

Au

Au

Au

Au

2 4 6 8 10

Full Scale 705 cts Cursor: 2.280 keV (31 cts) keVFull Scale 705 cts Cursor: 2.280 keV (31 cts) keVFull Scale 705 cts Cursor: 2.280 keV (31 cts) keV

Spectrum 1

71

Figure 3.5 UV-Vis spectra showing cumulative adsorption measurements observed over 24 hrs for the adsorption of Pt-DEN on 4 % NCNTs.

200 300 400 500 600 700 800

A

bso

rba

nce

/ a

.u.

Wavelength / nm

72

absorbance exists over the 24h period. This corresponds to the adsorption of

virtually all Pt-DENs from solution on the NCNT surface. The appearance of a

broad shoulder at 260 nm has been attributed to the absorbance at unreduced Pt2+

that remains bound to the exterior amine groups of the dendrimers.27 Figure 3.6

shows a picture depicting a visual representation of the Pt-DEN adsorption

process. Vial A contained a 20 µM solution of Pt-DENs which were dark brown

in color consistent with the presence of colloidal Pt in solution. Vial B contained a

1 mg/ ml of 4 % NCNTs dispersed in the Pt-DEN solution. The NCNTs seem

well dispersed in the solution showing no visual evidence of agglomeration. Vial

C represents the NCNT/ Pt-DEN solution after 24h of stirring the NCNT/ Pt-DEN

solution. The supernatant is transparent, consistent with the adsorption of Pt-

DENs from the solution onto the NCNT surface.

The adsorption behavior of the G4-NH2-terminated Pt-DENs for undoped

CNTs, 4 % NCNTs and 7.5 % NCNTs corresponding to different initial bulk Pt-

DEN solution concentrations for a 24 h period is shown in Figure 3.7. Adsorption

experiments were conducted in triplicate with Pt-DENs in the concentration range

of 5-20 µM. For each concentration, there was an increase in the percentage

adsorbed with increased immersion time. The extent of uptake of Pt-DEN

gradually decreases, finally reaching a limiting value as indicated by the plateau.

As discussed in section 3.2.4, UV-Vis data was used to construct adsorption

isotherms and estimate adsorption parameters by fitting to a Langmuir based

model for the adsorption of Pt-DENs. All of the curves in Figure 3.7 have a

similar shape, only differing in the amount of Pt-DEN adsorbed. The adsorption

73

Figure 3.6 Picture representing the Pt-DEN adsorption process on NCNTs. A) 20 µM Pt-DENs. B) Pt-DENs with NCNTs suspended in solution. C) NCNT/ Pt-DEN suspension after 24 hrs showing all DENs having been adsorbed onto the NCNTs rendering the solution colorless.

A B C

74

0 5 10 15 20

0.00

0.03

0.06

0.09

0.12

0.15

0.18

0.21

nm (µmol/g) K (L µmol

-1)

1. 0.37 ± 0.07 8 ± 1

2. 0.34 ± 0.03 6 ± 1

3. 0.29 ± 0.02 3 ± 1

P

t D

EN

ad

so

rbe

d / µ

mo

lg-1

Pt DEN Solution Concentration / µM

7.5 % NCNT (1)

4 % NCNT (2)

Undoped CNT (3)

Figure 3.7 Adsorption isotherms for G4-NH2 Pt-DEN adsorption on undoped CNT and NCNT supports.

75

of Pt-DENs onto 7.5 % NCNTs was almost twice as fast as that for undoped

CNTs for the same immersion time, especially within the first hour. Adsorption of

the Pt-DENs always went to completion for these specific carbons. The solutions

were transparent after 24 h, and showed no UV-Vis absorbance between the 250-

800 nm range. It should be noted the limited solubility of Pt-DENs in aqueous

solution (ca. 20 µM) restricts the range of the adsorption isotherm studies and

therefore complete saturation behavior for the adsorption isotherms was not

observed. However, trends in the relative characteristics of Pt-DEN adsorption on

the various CNT supports are easily discerned.

Pt-DEN adsorption values and adsorption equilibrium constants for 4 %

NCNTs, 7.5 % NCNTs and undoped CNT/ Pt-DEN composites calculated from

the isotherms are reported in the inset of Figure 3.7. The adsorption of the G4–

NH2 Pt-DENs was attributed to the strong van der Waals interactions 34 with edge

plane sites at NCNTs. One of the factors that hinders the adsorption of –NH2

terminated Pt-DENs on the undoped CNTs is the relative difficulty of wetting the

CNTs in solution. NCNTs on the other hand are easily wettable due to the

presence of positively charged nitrogen groups on the surface and to the existence

of more edge plane sites.18 These observations are supported by Raman

spectroscopic studies from Chapter 2 that estimate that 4 % NCNTs have roughly

2.4 times more edge plane sites per unit length compared to the undoped CNTs,

which is consistent with the 2.6 times greater adsorption affinity estimated from

the Pt-DEN adsorption isotherm data.

76

Similar studies on the adsorption of aqueous metal ion (Pb2+, Cd

2+, Cu

2+,

Ca2+ and Hg

2+) species on oxygen and nitrogen functionalized activated carbons

conducted by Xiao et al35 found strong correlations between the concentration of

nitrogen groups and metal ion adsorption from solution. While the adsorption

mechanism is still a topic of considerable discussion, the Pt-DENs remained

anchored to the carbon support and withstood repeated sonication and washings

with deionized water. The ability to load catalysts directly from solution onto

unmodified NCNTs circumvents a majority of traditional pre-processing steps

that modify the carbon and the catalyst and provides advantages in utilizing

properties of the pristine carbon and the catalyst.

3.3.3 TEM Analysis of DEs Adsorbed on CTs

TEM analysis was performed on the NCNT/ DEN composites to confirm

structural and compositional properties of the DENs on the NCNTs. Figure 3.8

shows a representative TEM image of Pt-DENs adsorbed from solution onto 4 %

NCNTs. The Pt nanoparticles themselves are stable inside the dendrimer template

and appear suspended a few angstroms above the carbon support. It is clear that

the DENs are well dispersed throughout the NCNTs and do not show any signs of

agglomeration. No evidence of agglomeration was found on undoped CNTs and

NCNTs with other nitrogen contents either. The inset in Figure 3.8 shows a high

resolution detail of the Pt particles on the NCNT surface. The presence of lattice

structures at the Pt nanoparticles shows that they are crystalline. Particle size

distributions calculated from the Pt-DENs adsorbed on the NCNTs show that they

77

Figure 3.8 TEM image of G4-NH2 Pt-DENs adsorbed on 4 % NCNT supports (Scale bar is 20nm). The inset shows high resolution structure of Pt nanoparticles (Scale bar is 5nm).

78

are virtually the same when compared to distributions calculated from as-

synthesized particles.

Figure 3.9 shows representative TEM images of Pd and PdAu DENs

adsorbed on 4 % NCNTs. The Pd DENs are found to be similarly well dispersed

and found to retain the same size as that of the as synthesized DENs. No

agglomeration of particles was observed. The PdAu DENs adsorbed on the 4 %

NCNTs had similar characteristics as that of the monometallic Pt and Pd-DENs.

EDS analysis of the adsorbed PdAu particles showed no changes in the

composition. A closer analysis of the left edge at the NCNT in Figure 3.9 B

corroborates evidence of the DENs suspended a few angstroms above the NCNTs.

Prolonged immersion of the NCNTs in DEN solutions were not found to affect

their structure as derived from TEM observations.

3.3.4 Comparison of Adsorption Characteristics of –H2 and –OH Terminated DEs

To further elucidate the difference in adsorption and the significance of

the terminal –NH2 groups at the dendrimers, experiments were carried out with

G4-OH Pt-DENs in the same concentration range. Pt-DENs synthesized using

these different dendrimer varieties were found to have highly comparable particle

size distributions. Figure 3.10 shows representative TEM images of Pt DENs

adsorbed on 4 % NCNTs from 20 µM solutions of G4-NH2 Pt-DEN and G4-OH

Pt-DENs. While the G4-NH2 Pt-DENs were well dispersed on the NCNTs, the

G4-OH Pt-DENs were found to agglomerate on the surface.

Reduced affinity was also seen for the –OH terminated DENs and was

attributed to weaker interactions between the alcohol groups and the carbon

79

Figure 3.9 TEM images of DENs adsorbed on the NCNT surface. a) Pd DENs adsorbed on 4 % NCNTs. b) PdAu DENs adsorbed on 4 % NCNTs

20 nm20 nm

a) b)

80

Figure 3.10 Representative TEM images comparing the adsorption of –NH2 terminated and –OH terminated Pt DENs on 4 % NCNTs. a) G4-NH2 Pt DENs adsorbed on 4 % NCNTs. b) G4-OH Pt DENs adsorbed on 4 % NCNTs

20 nm10 nm

a) b)

81

support. 4 % NCNTs suspended in a 20 µM solution of G4-OH Pt-DEN did not

seem to adsorb the Pt-DENs at the same rates as that of the –NH2 terminated Pt-

DENs. The supernatant after 24h had significant absorbance at 350 nm consistent

with considerable concentration (>40 % of initial concentration) of the G4-OH Pt-

DENs left unadsorbed in solution. These observations are consistent with

previous reports that G4-NH2 dendrimers adsorb more strongly than G4-OH

dendrimers onto supports including HOPG, gold and glassy carbon.27,31,34

3.3.5 Thermo Gravimetric Analysis of Pt DE loading on CTs

Total Pt loadings (wt. %) were quantified by centrifuging undoped CNT/

Pt-DEN and NCNT composites, decanting the supernatant and drying in air prior

to TGA analysis. The mass of iron catalyst used for CNT growth (~12 wt. %),

was constant for a particular variety of CNTs and was subtracted from the total

mass of the residue obtained after each TGA run.

TGA analysis performed on CNT samples with varying immersion times

in Pt-DEN solutions enabled control of the Pt loading with respect to immersion

times. Figure 3.11 shows TGA heating curves for control 7.5 % NCNT and 7.5 %

NCNTs with maximum Pt loadings. Maximum loadings were achieved by

replenishing the initial 20 µM Pt-DEN solution with a fresh solution after

complete adsorption of the DENs on NCNTs. This solution was allowed to stir

until no appreciable change was seen in the absorbance at 350 nm. Pt loadings as

high as 26 wt. % were observed using G4-NH2 terminated Pt-DENs on 7.5%

NCNTs after 30 h immersion. In contrast, loadings of only 15 wt. % were

82

Figure 3.11 Representative TGA heating curves for blank NCNTs and Pt-DENs adsorbed on 7.5 % NCNTs. Mass loading of Pt on NCNTs is calculated by subtracting final wt. % of NCNT/ Pt-DEN from final wt. % at blank NCNT.

0 200 400 600 800 1000

20

40

60

80

100

120

We

igh

t %

Temperature 0C

7.5 % NCNT / Pt-DEN

Control 7.5 % NCNT

83

achieved using G4-NH2 Pt-DENs on undoped CNTs and loadings of only ~7 wt.

% were achieved using G4-OH Pt-DEN on 7.5% NCNT after 30 h.

These observations are consistent with Raman data (Chapter 2) that

estimate higher edge plane content at NCNTs. The presence of a higher number of

edge plane sites effects higher adsorption rates and thus a higher mass loading of

DENs at the NCNTs as discussed in section 3.3.2.

3.3.6 Electrochemical Analysis of CT/ DE Composites

3.3.6.1 Cyclic Voltammetry Studies at CT/ Pt-DE Composites

The catalytic activity for ORR of the CNT/ Pt-DEN and NCNT

composites was studied using cyclic voltammetry conducted in a standard single

compartment three electrode electrochemical cell. To provide a direct basis for

comparison, voltammetric studies of CNT/ Pt-DEN and NCNT/ Pt-DEN

composites were performed on materials possessing the same Pt loading of 18±1

µg calculated based on a 5 µL aliquot of the composite obtained from a slurry

prepared with 1 mg CNT, or 1 mg NCNT, 75 µL of 0.15 wt. % Nafion in

anhydrous ethanol and 75 µL of 18 MΩ cm water. The aliquot was then drop cast

onto a glassy carbon electrode with a constant geometric surface area. Control

studies were also performed on undoped CNTs and NCNTs without Pt-DENs

present by casting films in a similar fashion to distinguish the effect of the carbon

support on ORR. Importantly, we were able to control the amount of Pt-DEN

dispersed on the CNT and NCNT supports by regulating the immersion time. The

same slurry preparation procedure was used to prepare a Pt/Vulcan carbon

standard composite (Johnson Matthey) with a Pt loading of 20 wt. %.

84

The peak potential (Ep) corresponding to CV’s for ORR at 7.5 % NCNT/

Pt-DEN was 730 mV more positive than the Ep seen for the undoped CNTs at -

210 mV as shown in Figure 3.12. This indicates that the Pt-DENs adsorbed on the

NCNT surface are in electrical contact with the carbon and are active for ORR. It

should be noted that the CV corresponding to ORR at 7.5 % NCNT/ Pt-DEN

showed the most positive shift in Ep value for an NCNT/ Pt-DEN composite.

Electroactive surface areas (ESAs) for the undoped CNT and NCNT/ Pt-DEN

composites were assessed by integration of the amount of charge associated with

the hydrogen adsorption on the Pt-DEN catalysts. The ESAs were 17.1 m2g-1 for

undoped CNT/ Pt-DEN, 20.6 m2g-1 for 4 % NCNT/ Pt-DEN and 24.9 m

2g-1 for

7.5 % NCNT/ Pt-DEN as shown in Table 3.1. Comparative analysis of the surface

area at the control CNTs and the ESA’s corresponding to adsorbed Pt at the CNTs

does not show correlations between those parameters consistent with findings

discussed in Chapter 2.

3.3.6.2 Rotating Disk Electrode Studies

Figure 3.13 displays ORR polarization curves for undoped CNT/ Pt-DEN

and various NCNT/ Pt-DEN composites in O2-saturated 0.1 M H2SO4 obtained

using an RDE at 1600 RPM.33 The performance of the as prepared CNT/ Pt-DEN

composites is consistent with previous voltammetry and RDE studies of Pt-DENs

immobilized on glassy carbon electrodes.36,37,38

However, the activity of Pt-DENs

supported on NCNT supports is higher than Pt-DENs supported on undoped

CNTs. Furthermore, the onset potential for ORR and activity of Pt-DENs

supported on NCNTs tracks with the amount of incorporated nitrogen, as

85

0.9 0.6 0.3 0.0 -0.3

0.0

0.1

0.2

Cu

rre

nt / m

A


Undoped CNT

3.7 % NCNT - Ni

4.0 % NCNT - Fe

7.5 % NCNT - Fe

7.5 % NCNT / Pt-DEN

Figure 3.12 Cyclic voltammograms for ORR at control undoped CNTs and NCNTs synthesized using Fe and Ni precursors compared to NCNT/ Pt-DEN composites in oxygen saturated 0.1 M H2SO4. ν = 20 mVs

-1.

87

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

4

3

2

1

0

1. Control Undoped CNT

2. Control 4% NCNT

3. Undoped CNT / Pt-DEN

4. 4 % NCNT / Pt-DEN

5. 5 % NCNT / Pt-DEN

6. Pt on Vulcan (20 wt.%)

7. 7.5 % NCNT / Pt-DEN

Cu

rre

nt

de

nsity / m

A c

m-2


1

2

3

4

5

6

7

Figure 3.13 Polarization curves for the ORR on G4-NH2 Pt-DEN/CNT and NCNT composites supported on a glassy carbon electrode immersed in an O2

saturated 0.1M H2SO4 solution. In all cases the Pt loading is 18±1 µg. Also shown are polarization curves for CNT and NCNT supports. Rotation rate =1600 rpm, scan rate =20 mV s

-1.

88

illustrated in Figure 3.13 for Pt-DENs supported on 4 %, 5 % and 7.5 % NCNTs.

We attribute the increase in catalytic activity at NCNTs to an improved carbon-

catalyst binding and increased electrical conductivity and also hypothesize that a

synergistic support effect is present with the NCNTs as they are known to

decompose reactive intermediates such as hydrogen peroxide into oxygen during

ORR.18 Furthermore, the mass transport limited current densities and mass

activities (estimated from RDE curves at +350 mV) for 7.5 % NCNT/ Pt-DEN

composites were 2.3 mA cm-2 and 0.05 mA g

-1, respectively. These values are on

par with reported values of 3.4 mA cm-2 and 0.085 mA g

-1 for conventional Pt

catalysts in the 2 nm size range.36 The quasi ideal electrocatalytic response for

NCNT/ Pt-DEN composites is attributed to improved mass transport and

electronic interactions of the Pt-DENs with the NCNT support. Clearly the CV

and RDE studies indicate that the Pt-DENs are catalytically active and function as

active centers for ORR.

89

3.4 COCLUSIOS

In summary, a facile method for preparing catalytically active carbon

supported Pt catalysts is demonstrated. The choice of the PAMAM G4-NH2

dendrimer template and terminal amine functional groups provides for uniform

preparation of size monodisperse catalysts and facilitates the controlled dispersion

and loading of the catalysts onto NCNT supports with well controlled structural

and compositional properties. The immersion based loading of catalysts onto a

carbon support by spontaneous adsorption to achieve specific Pt loadings offers a

less aggressive processing approach for preparing carbon supported catalysts

compared to the harsher catalyst dispersion and loading methods that generally

require oxidative treatment of the carbon support and/or chemical reduction of

metal salts to achieve selective binding of the active metal catalyst. Additionally,

the Pt-DENs and NCNTs serve as well defined models whose properties can be

controlled synergistically to achieve better performance at these composites as

depicted in Scheme 3.2. A synergistic activity is envisioned where the NCNT

support is reactive and serves to reduce the peroxide formed as a byproduct

during oxygen reduction at the metal catalyst. Studies of this nature will foster

better understanding of role of the carbon support on catalytic efficiency, catalyst

utilization and catalyst stability.

90

Scheme 3.2 Representation of synergistic ORR activity envisioned at NCNT/ Pt-DEN composites.

O2 OH

-

O2 + H2O2

OH-

+ H2O2

O2 OH-

91

3.5 REFERECES

1. Radovic, L. R.; Rodriguez-Reinoso, F. In Chemistry and Physics of Carbon; Thrower, P. A. Ed. Marcel Dekker, 1997, Vol. 25, p 243-358.

2. Wildgoose, G. G.; Banks C. E.; Compton, R. G. Small 2006, 2, 182.




6. Li, X.; Hsing, I. -M. Electrochim. Acta 2006, 51, 5250.

7. Yoon, B.; Wai, C. M. J. Am. Chem. Soc. 2005, 127, 17174.

8. Xing, Y. J. Phys. Chem. B 2004, 108, 19255.

9. Serp, P.; Hierso, J. C.; Feurer, R.; Kihn, Y.; Kalck, P.; Faria, J. L.; Aksoylu, A. E.; Pacheco, A. M.; Figueiredo, J. L. Carbon 1999, 37, 527.

10. Ijima, S. "ature 1991, 354, 56.

11. Dai, H. Surface Sci. 2002, 500, 218.

12. Li, W.; Liang, C.; Zhou, W.; Qiu, J.; Zhou, Z.; Sun, G.; Xin, Q. J. Phys. Chem. B 2003, 107, 6292.

13. Thess, A.; Lee, R.; Nikolaev, P.; Dai, H.; Petit, P.; Robet, J.; Xu, C.; Lee, Y. H.; Kim, S. G.; Rinzler, A.; Colbert, D. T.; Scuseria, D.; Tomanek, J. E.; Fischer, J. E.; Smalley, R. Science, 1996, 273, 483.

14. Serp, P.; Corrias, M.; Kalck, P. Appl. Catal. A: Gen. 2003, 253, 337.

15. Ebbesen, T. W. J. Phys. Chem. Solids 1996, 57, 951.

16. Lordi, V.; Yao, N.; Wei, J. Chem. Mater. 1998, 10, 718.

17. Yu, R.; Chen, L.; Liu, Q.; Lin, J.; Tan, K. –L.; Ng, S. C.; Chan, H. S. O.; Xu, G. –Q.; Hor, T. S. A. Chem. Mater. 1998, 10, 718.


92


20. Scott, R. W. J.; Wilson, O. M.; Crooks, R. M. J. Phys. Chem. B 2005, 109, 692.

21. Hirsch, A. Angew. Chem., Int. Ed. 2002, 41, 1853.

22. Guldi, D. M.; Rahman, G. M. A.; Zerbetto, F.; Prato, M. Acc. Chem. Res. 2005, 38, 871.

23. Niu, Y.; Crooks, R. M. C. R. Chimie 2003, 6, 1049.

24. Planeix, J. M.; Coustel, N.; Coq, B.; Brotons, V.; Kumbhar, P. S.; Dutartre, R.; Geneste, P.; Bernier, P.; Ajayan, P. M. J. Am. Chem. Soc. 1994, 116, 7935.

25. Li, W.; Liang, C.; Qiu, J.; Zhou, W.; Han, H.; Wei, Z.; Sun, G.; Xin, Q. Carbon 2002, 40, 787.

26. Satishkumar, B. C.; Vogl, E. M.; Govindaraj, A.; Rao, C. N. R. J. Phys. D: Appl. Phys. 1996, 29, 3173.

27. Ye, H.; Scott, R. W. J.; Crooks, R. M. Langmuir 2004, 20, 2915.

28. Zhao, M.; Crooks, R. M. Angew. Chem., Int. Ed. 1999, 38, 364.

29. Zhao, M.; Crooks, R. M. Adv. Mater. 1999, 11, 217.

30. Scott, R. W. J.; Ye, H.; Henriquez, R. R.; Crooks, R. M. Chem. Mater. 2003, 15, 3873.

31. Scott, R. W. J.; Wilson, O. M.; Oh, S. –K.; Kenik, E. A.; Crooks, R. M. J. Am. Chem. Soc. 2004, 126, 15583.

32. Vijayaraghavan, G.; Stevenson, K. J. Langmuir 2007, 23, 5279.

33. Schmidt, J.; Gasteiger, H. A; Stab, G. D.; Urban, P. M.; Kolb, D. M.; Behm, R. J. J. Electrochem. Soc. 1998, 145, 2354.

34. Sun, L.; Crooks, R. M. Langmuir 2002, 18, 8231.

35. Xiao, B.; Thomas, K. M. Langmuir 2005, 21, 3892.

36. Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl. Catal. B 2005, 56, 9.

93

37. Paulus, U. A.; Schmidt, T. J.; Gasteiger, H. A.; Behm, R. J. J. Electroanal. Chem. 2001, 495, 134.

38. Mayrhofer, K. J. J.; Blizanac, B. B.; Arenz, M.; Stamenkovic, V. R.; Ross, P. N.; Markovic, N. M. J. Phys. Chem. B 2005, 109, 14433.

94

CHAPTER 4

Metal Organic Chemical Vapor Deposition of anocarbon Supported Mono- and Bimetallic Catalysts for

Oxygen Reduction

4.1 ITRODUCTIO

The efficiency and utilization of heterogeneous catalysts is directly

dependant on the synthetic protocols involved in the loading and subsequent

activation of the catalyst. These steps are typically refined to optimize size,

composition, dispersion and surface area of the catalyst on the carbon support.

While a host of liquid phase methods like microemulsion1 and impregnation

2 are

used to deposit noble metal catalysts on carbon supports, they typically require

harsh chemical or physical processing steps of the carbon support to facilitate

catalyst loading and dispersion. Typical procedures used for activating carbon

supports and creating surface oxygen functionalities use strong acids3 such as

HNO3 or H2SO4 or strong oxidizing agents.4 These methods alter the intrinsic

properties of the support and tend to be variable and time consuming processes.

The use of the above mentioned procedures can also hinder understanding of the

role of the carbon support properties on specific catalytic activity. To this end, we

have been studying the specific role of nitrogen doped carbon nanotube (NCNT)

supports in technologically important reactions such as ORR as dealt with in

Chapter 2. Previous reports from the Stevenson research group have demonstrated

that NCNTs show inherent catalytic activity for both ORR and the heterogeneous

95

decomposition of hydrogen peroxide.5, 6

While as grown NCNTs are less active

than Pt supported on carbon supports, we speculate that they promote a catalytic

synergism by scavenging intermediates from ORR such as hydrogen peroxide.5

The removal of these reactive intermediates is a key step in improving the

catalytic activity of a support/ catalyst combination and also leads to supports that

are better resistive to corrosion7 over extended periods. These properties of the

NCNTs has been taken advantage of in previous studies that pair these supports

with catalysts synthesized using dendrimer templates,8 as discussed in Chapter 3.

Although the NCNT/ dendrimer composites were active catalysts, their

electrochemical surface areas were lower compared to commercial catalysts. The

cost of the dendrimer templates also hinders commercialization aspects of this

particular scheme for preparing templated nanoparticles and alternate routes are

necessary to prepare cost effective active carbon/ catalyst composites.

Literature reports have dealt with the surface selective nature of CVD for a

variety of metals such Pt9, Pd

9, Rh

10 and Ag

11 using metal organic chemical vapor

deposition (MOCVD) precursors. The influence of surface area, porosity and the

presence of anchoring sites at specific supports on the preparation of supported

catalysts using MOCVD routes have also been discussed.12 A number of different

techniques have been employed to effect the CVD process and a few hypotheses

exist as to the nucleation and growth mechanism. Nuzzo et al. have performed

extensive studies on the CVD of hexafluoroacetylacetonate metal precursors on

various substrates and have proposed a fairly comprehensive mechanism.13,14

A

combination of XPS, RAIRS and TPD studies on the CVD of

96

bis(hexafluoroacetylacetonate)-palladium(II) Pd(hfac)2 on a copper surface has

shown that this process involved several principal steps. The first was the

adsorption of Pd(hfac)2 molecules on the Cu surface at the appropriate

sublimation temperatures. The second step involved the reduction of PdII to Pd

0

by the Cu surface. Processes that involve CVD on a non metallic surface have

effected this step by employing a reductive gas such as H2 during the

deposition.15,16

The use of a reductive gas flow or a reductive surface, results in

the dissociation of the hfac ligands from the Pd that constitutes the third step. The

hfac ligands were seen to react with the surface Cu to form Cu(hfac)2 which was

then desorbed from the surface. In the case of CVD at a non metallic surface, the

decomposition and removal of the precursor ligands was seen to be facilitated by

the H2 gas during reduction of the metal.

A majority of the CVD processes are applied for the deposition of thin

films on relatively smooth surfaces.17 Deposition of particles through CVD has

been typically nucleated by the creation of aberrations and induction of specific

functionalities on the surface of the support by aggressive processes as discussed

in the preceding paragraphs. Several groups have studied the dependence of the

nucleation rate of metals on the nature of the support and have agreed in principle

that surfaces customized with specific composition and structure can be used to

control the nucleation step of CVD facilitating tailored growth of

nanoparticles.13,16

Traditional CVD processes employed towards the deposition of

nanoparticle catalysts on a carbon support utilize a fluidized bed.18,19

Fluidized

97

bed processes use a gas stream flowing from the bottom of a reactor to suspend

highly functionalized carbon supports in the deposition zone. This enables better

interaction of the precursor with the carbon in the deposition zone resulting in a

fairly uniform dispersion of the metallic nanoparticles on the carbon surface.

The ability to manipulate in-situ,5,6 the structure, composition, surface

area and the amount of edge plane sites on NCNTs provides a significant

advantage over traditional carbons for use as supports in the CVD scheme

because there is no further need to modify the surface to facilitate catalyst

loading. The capability to synthesize NCNTs with a remarkable degree of

alignment and uniformity on a variety of substrates facilitates the direct

deposition of metal nanoparticles. Substrates with the as synthesized NCNTs can

then be directly exposed to the CVD precursors for controlled and uniform

nanoparticle growth without necessitation of a fluidized bed reactor.

The attractive properties of NCNTs along with the ease of tailoring and

availability of a variety of catalyst precursors used in MOCVD provides

significant advantages over other schemes for direct synthesis of support/ catalyst

combinations without extensive pre- and post processing steps. Since MOCVD

precursors can serve as sources for both the synthesis of the carbon support and

the ORR metal catalyst, it should be possible to create active catalysts directly on

high-surface area carbons in fewer processing steps. This chapter details the

loading of monodisperse Pt, Pd and PtPd catalysts in a subsequent step after the

synthesis of NCNTs. However, it is also possible to synthesize carbon/ catalyst

combinations from a single precursor source- a topic that is currently being

98

explored at the Stevenson research lab. The MOCVD route offers promise for the

direct dispersion and activation of mono- and multimetallic ORR catalysts on

carbon supports and eliminates the current inevitable problems involving loading,

sintering and activation steps associated with traditional solvent based catalyst

preparation schemes.

While this chapter discusses the synthesis aspects of monometallic Pt, Pd

and bimetallic PtPd nanoparticles on nanocarbons in detail, emphasis is placed on

the characterization of the bimetallic PtPd catalysts. Although Pt catalysts show

high activity for ORR, their performance decreases over time due to the formation

of Pt-OH20 that inhibits adsorption and reduction of oxygen on the catalyst

surface. Recent studies by the Adzic group20 conducted on carbon supported

monometallic Pt (Pt/C) and bimetallic PtAu (PtAu/C) have shown that the Pt/C

catalyst loses over 45 % of its electroactive surface area after 30,000 cycles (0 –

1.2 V vs. NHE) in oxygen saturated 0.1 M HClO4. Under the same experimental

conditions the bimetallic PtAu/C retained its original electroactive surface area.

Adzic et al attributed the stability of the PtAu bimetallic catalyst to the decreased

oxidation of the Pt nanoparticles covered by the Au surface. Multimetallic Pt

catalysts with synergistic components that help prevent oxide formation on Pt are

of vital importance in improving ORR efficiency.

99

4.2 Experimental

4.2.1 Synthesis of Undoped CTs and CTs

CNTs were prepared using the floating catalyst chemical vapor deposition

method5,6 described in chapter 2. Briefly, xylene (Aldrich) as carbon source and

ferrocene (Aldrich) as catalyst were used for the growth of the undoped CNTs

while a pyridine (Fisher) and ferrocene precursor combination was used for the

synthesis of nitrogen-doped CNTs. The use of a pyridine precursor allowed for a

controlled doping of ~ 4 at. % N on the NCNT surface. Higher surface nitrogen

concentrations (5-10 at. % N) were obtained by introducing a regulated stream of

ammonia gas (Aldrich) into the CVD furnace system along with the pyridine and

ferrocene precursors.

4.2.2 CVD of Mono- and Bimetallic anoparticles on CTs

Monometallic Pt, Pd and bimetallic PtPd catalyst nanoparticles were

directly deposited via CVD in a sequential step on the as grown CNTs using

thermally labile precursors. Once the synthesis step for a particular variety of the

CNTs was completed, a carefully measured mass of Pt(acac)2 or Pd(acac)2 (Alfa

Aesar, 99.9%) was weighed and placed in a ceramic boat in zone 1 of the furnace

for deposition of Pt or Pd nanoparticles respectively. While the gas flow stream

comprised of 400 sccm Ar and 70 sccm H2, zone 1 was maintained at 170 C (190

C for Pd) and zone 2 was maintained at 220

C for sublimation and subsequent

decomposition respectively of the Pt or Pd catalysts on the CNTs. The

sublimation and decomposition process spanned a 20 minute period and a dark

100

organic mass was left in the ceramic boat. After the furnaces were allowed to cool

to room temperature, the catalysts were retrieved from the interior of the quartz

tube and stored in airtight vials prior to characterization.

In the case of bimetallic PtPd nanoparticles, a 50:50 mass ratio of

Pt(acac)2 and Pd(acac)2 was weighed and placed in a ceramic boat at zone 1 of the

furnace system. The carrier gas stream consisted of 400 sccm Ar and 70 sccm H2.

Zone 1 was maintained at 190 C and zone 2 at 220

C for sublimation and

decomposition respectively of the catalyst precursors.


Transmission electron microscopy (TEM) characterization of the CNT

supported catalysts was performed on a JEOL 2010F microscope operating at 200

kV. Energy dispersive spectroscopy was performed on an Oxford instruments

INCA detector that was part of the TEM instrumentation. TEM samples were

prepared by dispersing the catalysts in anhydrous ethanol and drop casting on a

Cu TEM grid covered with lacey carbon film.

4.2.4 Raman Characterization

A Renishaw inVia system equipped with a 514.5 nm Ar laser at 3 mW/

cm2 and 100X aperture was used. Spectra were scan averaged for 300 s. Bands at

1220, 1351, 1487, 1583 and 1624 cm-1 corresponding to I, D, D′′, G, and D′ bands

denoted by Cuesta et al13 were fit using a Peak Fit 4 software package to

correlation factors greater than 0.998. A linear baseline correction was used to

compensate for photoluminescence background.

101

4.2.5 X-Ray Photoelectron Spectroscopy Characterization

X-ray photoelectron spectroscopy characterization of the samples was

performed using a PHI 5700 ESCA system operating at a scan step size of 0.1 eV

and an Al Kα monochromatic line (1486.6 eV) calibrated with Au 4f7/2, Ag 3d5/2

and Cu 2p3/2 signals. All spectra were scan averaged 5 times. Atomic percentages

were quantified from survey scans relative to carbon, iron and nitrogen. FITT 1.2

(Photoelectron Spectroscopy lab, Seoul National University) software with

Shirley background corrections was used to analyze the spectra.



7000 analyzer. Samples (~5 mg) were held in platinum pans heated to 800 C at 5

C / min in flowing air (Praxair, 99.998%).


Electrochemical measurements were carried out on a EG&G Instruments



on a Corrware (Scribner Associates) software package. Sample slurries prepared

with 0.15 wt. % nafion in anhydrous ethanol and de-ionized water were drop cast

as a film on a glassy carbon RDE (0.5 cm diameter, Pine Instruments) polished to

a mirror finish and allowed to dry under a stream of Ar (Praxair 99.98%) prior to

electrochemical measurements. Cyclic voltammograms were obtained in a

standard three electrode cell with a gold counter electrode and a Hg/Hg2SO4

reference electrode in O2 (Praxair) saturated 0.1 M H2SO4 prepared with

102

deionized water (>18 MΩ cm). All potentials were converted to NHE for

comparison to literature values. For electro-active surface area measurements, CO

(Praxair) was dosed into the electrolyte solution for 20 min while the electrode

was held at a constant potential and the solution then purged with Ar for 30 min to

remove CO from the bulk.

Prior to oxygen reduction and CO stripping experiments, the CNT/catalyst

film was cycled between 0.8 and -0.2V in order to make the film hydrophilic and

achieve steady state voltammograms.

Chronoamperometry experiments were carried out in CO saturated 0.1 M

H2SO4. CO was dosed into the electrolyte for 10 min while the electrode was held

at 0.11V. CO in bulk was purged by bubbling in a stream of Ar for 20 min.

Electrode potentials were subsequently stepped up and chronoamperometric

transients recorded on a 30 S time scale.

103


4.3.1 Morphological Properties of Catalysts Synthesized by MOCVD

The surface interaction between a catalyst precursor and the carbon

support is important in the synthesis of carbon supported catalysts through

chemical vapor deposition.12 The choice of specific precursors, the composition

and flow rate of the feed gas stream highly influences the size, crystallinity,

structure and composition of the catalyst particle. The presence of a reducing

agent such as hydrogen21 in the feed stream greatly reduces the deposition of free

carbon contaminations from the metal organic precursor that significantly

decreases the available surface area of the metal. The introduction of the catalyst

precursor into zone 1 of the furnace after the synthesis of CNTs in zone 2 was

carefully controlled and the quartz tube purged with Ar for 30 minutes to remove

trace oxygen. The presence of oxygen could lead to the formation of oxygen

functionalities on the surface of the CNT support, affect efficient decomposition

of the precursor and the electrochemical performance of the catalyst composite.

Initial experiments with the sublimation and decomposition of PtMe2(COD) on

CNT supports left substantial amounts of carbon impurities22 on the surface of the

CNT/catalyst composites. While the amount of carbon impurities was reduced

using appropriate flow rates of hydrogen in the feed stream, the particle sizes

were found to vary over a wide range (15-30 nm) and were not evenly distributed.

The use of Pt(acac)2 and Pd(acac)2 precursors allowed for a narrower size

distribution and smaller particle sizes compared to the PtMe2(COD) precursor

104

used for the Pt source. Figure 4.1 shows TEM images of Pt and Pd catalysts

directly deposited on NCNT supports without additional processing or post

activation steps. As shown in Figure 4.1a and the corresponding particle size

histogram in Figure 4.1b the Pt nanoparticles are fairly monodisperse with an

average diameter of 2.1 nm and appear to be dispersed uniformly over the NCNT

support. The Pt catalysts also appear to be strongly anchored with the NCNT

support and withstand repeated washing and sonication steps. In comparison Pd

catalysts prepared using a Pd(acac)2 precursor under similar conditions were of a

larger average particle size of 11.8 nm (Figure 4.1c and 4.1d).

Figure 4.2 shows a representative TEM image and the particle size

distribution of PtPd on 6.5 % NCNTs. Literature data12,21

suggest that the carbon

impurities from the precursor constitute less than 1% of the total mass consistent

with TEM observations. The inset in Figure 4.2 shows that PtPd bimetallic

particles deposited on 6.5% NCNTs were crystalline with an average particle size

of 3.1 nm. The graphitic nature of the carbon substrate is also clearly represented.

Overall the particles were fairly monodisperse with no evidence of

agglomerations. Average particle size and particle size distributions were

calculated from a total of at least 100 particles measured across three different

batches. The energy dispersive spectrum (Figure 4.2c) of single PtPd bimetallic

particles averaged across at least three different samples shows that the particles

consist of 62 ± 2 % Pt and 38 ± 3 % Pd.

105

Figure 4.1 TEM images of a) Pt and c) Pd catalysts on NCNT supports. Corresponding particle size histograms for b) Pt and d) Pd catalysts.

a)

c)

8 10 12 140

10

20

30

Counts

Particle size / nm

d)

1 2 3 40

10

20

30

40

Counts

Particle size / nm

b)

106

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.00

10

20

30

Counts

Particle size / nm

b)

Figure 4.2 TEM image of a) PtPd particles on 6.5 % NCNT supports. Inset shows high resolution image of a PtPd bimetallic catalyst particle. b) Corresponding particle size histogram for PtPd catalysts. c) Representative energy dispersive spectrum of a single PtPd nanoparticle.

a)

c)

107

4.3.2 Raman Analysis

First order Raman spectra of the catalysts supported on different varieties

of CNTs (Figure 4.3) were collected primarily to quantify and ensure that the

amount of edge plane sites on the CNT supports loaded with catalysts was

comparable to that of the pristine CNT varieties thereby minimizing ambiguity on

other factors that might influence catalyst loading. The amount of disorder on

each of the CNT supports were measured by fitting and integrating the intensities

of the D and G bands over an average of three different batches as shown by

Cuesta et al.23 1/ La values calculated for undoped CNTs, 4 % NCNTs and 6.5 %

NCNTs loaded with PtPd were 0.12, 0.3 and 0.52 nm-1 respectively and were

agreement to reported values. Raman analysis was found to be an accurate

indicator of the amount of edge plane content in the CNTs.

4.3.3 XPS Analysis

Figure 4.4a shows a representative XPS survey spectrum of the PtPd

catalysts on 6.5% NCNT. The carbon peak at 284.7 eV is close to values reported

earlier.5, 24

It should be noted that there are no remarkable peaks in the 286-289

eV shoulder region that indicate a significant presence of oxygen like

functionalities on the surface of the NCNTs, which lends credence to reasoning

that these groups do not influence the deposition of catalysts on the NCNTs. High

resolution spectra of the Pt and Pd regions are shown in Figure 4.4b and 4.4c

respectively. The Pt 4f region shows doublets arising from the spin-orbital

splitting of the 4f7/2 and the 4f3/2 states. Peaks at 71 and 74.5 eV are attributed to

metallic Pt (Pt(0)). The Pd region shows peaks at 335 and 341 eV that have been

108

400 800 1200 1600 2000

Normalized Counts

Raman Shift / cm-1

PtPd on Undoped CNT

PtPd on 4% NCNT

PtPd on 6.5% NCNT

Figure 4.3 Raman spectra of PtPd bimetallic catalysts on NCNT supports with varying surface concentrations of nitrogen.

109

65 70 75 80 85 90

Normalized Counts

Binding Energy / eV

High resolution

Pt spectrum71 eV

74.5 eV

b)

330 335 340 345 350

Normalized Counts

Binding Energy / eV

High resolution

Pd spectrum335.1 eV

340.5 eV

c)

Figure 4.4 XPS spectra of PtPd supported on 6.5 % NCNTs. a) Survey spectrum of PtPd showing Pt and Pd regions. b) High resolution XPS spectrum of Pt and c) Pd catalysts supported on 6.5 % NCNTs.

0 500 1000

PtPd

Survey spectrum

Normalized Counts

Binding Energy / eV

Pt

Pd

a)

110

attributed to zerovalent Pd. As reported by S.H.Y Lo at al,25 the 3d3/2 peak can be

fit to different oxidation states of Pd with peaks at 340.5, 341.4 and 342.6

corresponding to Pd, PdO and PdO2. The peak at 335 eV was also fit to the same

oxidation states with bands at 335.1, 336.2 and 337.2 corresponding to Pd, PdO

and PdO2. Peak fits for the PtPd catalysts were similar across the different

varieties of CNT supports analyzed. High resolution spectra of the nitrogen region

were compared to NCNT spectra without any catalyst loading and agreed with

reported literature data5 with respect to the position of the pyridinic, pyrrolic and

quaternary nitrogen and quantitation of the corresponding nitrogen moieties.

4.3.4 TGA Analysis

Quantitation of catalyst loading amounts on the CNTs was performed by

TGA. TGA analysis in conjunction with Raman spectral analysis of the amount of

edge plane content in CNTs provided strong correlations to the influence of the

edge plane sites on catalyst loading. For comparison purposes, data presented

correspond to catalysts loaded on undoped CNT, 4 % NCNT and 6.5 % NCNT

supports.

Figure 4.5 shows TGA curves of various CNTs that were subjected to the

same loading protocols utilizing the same precursor mass. Control undoped

CNTs, 4% NCNTs and 6.5% NCNTs studied at the same temperature range had

residual iron (Fe2O3) mass percentages of 7 ± 1 %, 9 ± 1 % and 12 ± 1 %

respectively. Subtracting the mass of iron from the final mass obtained at 800 C,

PtPd loadings of 0.9 ± 0.3 %, 4 ± 1 % and 18 ± 1 % were obtained for undoped

CNTs, 4% NCNTs and 6.5% NCNTs respectively. The difference in loading was

111

0 200 400 600 8000

20

40

60

80

100

Weight %

Temperature / oC

PtPd 6.5% NCNT

PtPd 4% NCNT

PtPd undoped CNT

Figure 4.5 TGA heating curves for PtPd catalysts on various CNT supports

112

0.4 0.8 1.2 1.6 2.0 2.4

0

3

6

9

12

15

18

PtPd loading / Weight %

ID / I

G ratio

6.5% NCNT

4% NCNT

Undoped CNT

Figure 4.6 Effect of edge plane content at NCNTs on PtPd loading as determined by TGA.

113

directly correlated to the amount of edge plane content in the respective CNTs as

shown in Figure 4.6. Careful control of TGA analysis conditions also provides

insight into burn off temperatures and rates for various CNTs that can be directly

correlated to the amount of disorder on the surface.23 The measured burn off rates

were slightly lower than pristine CNTs reported by Maldonado et al5 and were

consistent with literature reports.26


4.3.5.1 Cyclic Voltammetry

The electroactive surface area is an important benchmark in comparing

catalysts.27 The commonly accepted method of determining the electroactive

surface area (ESA) of a catalyst is the integration of charge associated with the

amount of hydrogen adsorbed/desorbed on the surface of the electrode. While this

method of determining the ESA is fairly accurate for Pt catalysts, it is difficult to

normalize the surface areas of non noble metal catalysts and in cases of Pt based

bimetallic catalysts. The adsorption/ desorption of hydrogen on the surface of

PtPd is known to be particularly dependant on the size and surface composition of

the catalyst.28 Voltammograms of PtPd catalysts of varying compositions have

been shown to have distinct differences in the ‘hydrogen region’ and subsequently

lead to significant differences in the surface areas calculated through the

integration of the charge associated with the adsorption/ desorption of hydrogen.

Figure 4.7 shows a typical cyclic voltammogram of the PtPd catalysts supported

on 6.5% NCNT. The voltammogram is similar to that of other PtPd catalysts on

the nanoscale as reported in literature.29 This voltammogram was also comparable

114

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 -0.2-2

-1

0

1

2

3

4

5

Current density / (mA/cm

2)


O2 sat. 0.1M H

2SO

4

20 mV/s

PtPd 6.5% NCNT

Figure 4.7 Representative cyclic voltammogram for PtPd catalysts supported on 6.5 % NCNTs in O2 saturated 0.1M H2SO4.

115

to pure Pd catalysts and the peak near 0 V has been ascribed to the combination of

hydrogen desorbed from the bulk of the metal with that of hydrogen at the

surface. The peak at 0.2 V has been ascribed to the desorption of hydrogen on the

surface and also in the α- phase of the Pd catalyst.30 Grden et al

29 and Eley et al

31

have shown that the appearance of two distinct peaks related to the desorption of

hydrogen suggest there are two distinct phases of adsorbed hydrogen that exist on

the Pd surface that affect the ESA. ESA values calculated from hydrogen

adsorption on PtPd supported at undoped CNTs, 4 % NCNTs and 6.5 % NCNTs

were 1.8 ± 0.6 m2/g, 13 ± 1 m

2/g and 56 ± 3 m

2/g respectively.

4.3.5.2 ESA Analysis by CO Stripping

With some of the inherent problems in using the hydrogen region to

calculate the ESA, CO adsorption and oxidation on the surface of catalysts is an

effective method that provides information on the ESA regardless of whether the

catalyst is composed of noble or non-noble metals and insight into surface

poisoning. Current-time transients with respect to CO oxidation at various

potentials also provide insight into the effects of particle size and mechanism of

CO monolayer oxidation especially with regard to OHads on the surface of the

catalysts.32 CO electro-oxidation was carried out on PtPd films drop cast on a

highly polished glassy carbon (GC) RDE and dried under argon to prevent

oxidation of the catalyst layer. CO stripping on PtPd catalysts deposited on 6.5%

NCNT is shown in Figure 4.8 The portion of the CO stripping voltammogram in

the hydrogen region was devoid of any features associated with the adsorption/

desorption of hydrogen indicating that the catalyst surface was saturated with CO.

116

1.2 0.9 0.6 0.3 0.0 -0.3

-3

0

3

6

9

12

O2 sat. 0.1M H

2SO

4

20 mV/s

Current density / mA cm

-2


CO stripping on PtPd 6.5 % NCNT

Subsequent voltammogram

Figure 4.8 CO stripping voltammogram on PtPd catalysts supported on 6.5 % NCNTs. CO was dosed into solution for 20 min followed by 30 min of Ar purge to remove CO in bulk.

117

Furthermore, it was also indicative of oxidation of CO on the surface and not

from the bulk solution. The peak potential and onset for the PtPd catalysts were

0.77 V and 0.65 V respectively and were found to be comparable to Pt catalysts in

the same size range.32 The feature around 0.5 V was attributed to the reduction of

surface oxide. The voltammogram recorded in succession to the CO stripping

showed peaks associated with the adsorption/ desorption of hydrogen and is

indicative of the regeneration of the surface after CO oxidation.

ESA values calculated from CO electro-oxidation experiments were

compared over three consecutive cycles and displayed no significant decrease in

the surface area of the catalysts. ESA values for the PtPd catalysts estimated by

hydrogen adsorption/ desorption and CO stripping are reported in Table 4.1. It

should be noted that the voltammogram recorded in succession to the CO

stripping voltammogram was used to calculate the charge associated with the

adsorption/ desorption of hydrogen. The differences in the surface areas

calculated from CO and hydrogen adsorption peaks are apparent from Table 4.1.

The ESA’s calculated from hydrogen adsorption/ desorption are greater than the

ESA’s calculated from CO oxidation because of errors associated with the

integration of the amount of charge due to hydrogen adsorption/ desorption. We

believe that surface areas calculated from CO stripping serve as more accurate

indicators for comparison to activity benchmarks.

4.3.5.3 Chronoamperometric Studies

Chronoamperometric transients for CO oxidation on the PtPd catalysts

were conducted to study the influence of the catalyst particle size on activity.

119

These experiments also provide insight into the mechanism of CO oxidation and

the effect of OHads on the surface.33 Figure 4.9 shows chronoamperometric

transients for the PtPd catalysts supported on 6.5% NCNT. These transients were

comparable to Pt particles of the same average size supported on GC as reported

in Maillard at al32 and show current maxima at faster time scales. This particular

aspect was attributed to the presence of smaller particles less than 0.8 nm that

result from the CVD process. The transients show well defined peaks for the

current rise and decay34 and the current maximum increases and shifts to faster

time scales corresponding to an increase in the oxidation potential. The rise and

decay of the current maxima is a subject of considerable argument due to

ambiguity in attributing multistep electrochemical processes and rate limiting

steps to particular characteristics of the curve, but it is generally agreed that the

formation of CO2 occurs during the initial current rise. As has been dealt with in

several literature sources35 transients were not background subtracted due to the

influence of CO coverage on surface oxidation and double layer charging.

4.3.5.4 RDE Studies

The activity of PtPd catalysts and the synergistic role of the nanocarbon

support were studied for ORR using RDE experiments. The carbon support plays

a vital role in enhancing the activity of the catalysts prepared by the MOCVD

route since the NCNTs are known to spontaneously decompose peroxide which is

a byproduct of ORR.36 ORR peak potentials were more positive for catalysts

supported on NCNTs than on undoped CNTs with identical catalyst loadings.8

RDE experiments facilitated calculations as to the efficiency of ORR on the

120

0 5 10 15 20 25 30

0.0

0.2

0.4

0.6

0.8

1.0

Current / mA

Time / s

0.7 V

0.8 V

0.85 V

Figure 4.9 CO stripping transients on PtPd catalysts supported on 6.5 % NCNTs. CO was dosed into solution for 20 min followed by 30 min of Ar purge to remove CO in bulk. The working electrode was held at 0.11V during CO dosing and raised to stripping potentials.

121

catalysts and compare activities to commercial Pt values in literature. RDE

polarization curves shown in Figure 4.10 for PtPd catalysts supported on 6.5%

NCNT were comparable to benchmark catalyst systems. The mass transport

limited current density at 1600 rpm37 was 2.5 mA/ cm

2 which was comparable to

commercial Pt catalysts on XC-72 in the same size range.27 Koutecky- Levich

plots derived from the RDE data indicate that 3.6 electrons were involved in ORR

which also compares well to published literature data.27 To account for the effects

of the bisulfate in sulfuric acid, RDE experiments were carried out in 0.1 M

HClO4. This resulted a significant decrease in the current density of a peak shaped

feature attributed to the bisulfate adsorbed on the catalyst surface as shown in

Figure 4.11.

4.3.5.5 Stability of PtPd Catalysts

One of the primary advantages of MOCVD is that it allows for the

synthesis of nanoparticles directly on the CNT surface without the need for

templates to control size distributions of the as synthesized nanoparticles. This

allows for direct interrogation of the synergistic effects of the substrate and the

catalyst devoid of possible interactions with templating agents or influence from

multiple synthesis and processing steps that are typically needed to design

catalysts of similar size distribution. While MOCVD is a ‘soft’ chemical route

that does not require heat treatment at high temperatures, the stability of these

catalysts are on par with commercial Vulcan based catalysts. Figure 4.12 shows

representative voltammograms for the 1st and the 50

th cycles for PtPd catalysts

supported on 6.5% NCNTs. It was seen that the oxygen reduction peak shifts

122

Figure 4.10 Polarization curves for ORR on PtPd catalysts supported on 6.5 % NCNTs in O2 saturated 0.1 M HClO4. The RDE was rotated between 250 – 3000 rpm in increments of 250 rpm. Scan rates were 20 mV /s.

1.0 0.8 0.6 0.4 0.2

0.0

0.2

0.4

0.6

0.8

1.0

O2 sat. 0.1M HClO

4

ν = 20 mV/s

Current / mA


123

1.0 0.8 0.6 0.4 0.2

0.0

0.2

0.4

0.6

0.8

1.0

Current / mA


O2 sat. 0.1M H

2SO

4

ν = 20 mV/s

Figure 4.11 Polarization curves for ORR on PtPd catalysts supported on 6.5 % NCNTs in O2 saturated 0.1 M H2SO4 showing the effects of adsorbed bisulfate. The RDE was rotated between 250 – 3000 rpm in increments of 250 rpm. Scan rates were 20 mV /s.

124

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 -0.2

-2

0

2

4

6

8

Current density / (mA/cm

2)


PtPd 6.5% NCNT 1st cycle

PtPd 6.5% NCNT 50th cycle

O2 sat. 0.1M H

2SO

4

20 mV/s

Figure 4.12 Extended cycling of PtPd catalysts supported on 6.5 % NCNTs in O2 saturated 0.1 M H2SO4.

125

positive by 12 mV for the PtPd catalysts. This was attributed to cleansing the

catalyst surface of impurities associated with the MOCVD process. A marginal

decrease in the current density of 0.18 mA/ cm2 for PtPd was also seen. This loss

in current density for PtPd was attributed to the possible rearrangement of the

catalyst surface. For the PtPd catalysts, the plateau observed between 0.55 and 0.7

V on the cathodic sweep was attributed to the reduction of PdO.38 Features in the

hydrogen region become more resolved over extended cycling while no

significant loss was seen with the amount of charge associated with hydrogen

adsorption/ desorption supporting the idea that the PtPd is electrocatalytic and

stable.

126

4.4 COCLUSIO

In conclusion, the MOCVD scheme for preparing mono- and bimetallic

catalysts directly on undoped and nitrogen doped carbon supports was shown to

compare well with catalysts prepared through other processes that require a

sequence of pre- and post synthesis steps. The ability to deposit fairly

monodisperse catalyst nanoparticles that bind strongly to the NCNT supports and

display synergistic effects on the oxygen reduction reaction was unambiguously

defined through a series of experiments. The tunability of the amount of edge

plane content on the nanocarbon substrate offers distinct advantages with regard

to preferential catalyst loading and decomposition of peroxide specific to ORR.

The MOCVD process is expected to serve as a model system for rapid and direct

assembly of carbon supports and varied catalyst compositions that provide

promising activity for oxygen reduction. This method of catalyst screening is

expected to be straightforward for a wide variety of combinations due to the

availability of thermally labile precursors for a range of different metals that show

promise for ORR activity either by themselves or in multimetallic combinations.

This could be taken advantage of for studies on catalyst dispersion, availability

and stability in combination with a variety of carbon supports.

127

4.5 REFERECES

1. Yoon, B.; Wai, C. J. Am. Chem. Soc. 2005, 127, 17174.





6. Maldonado, S.; Stevenson, K. J. J. Phys. Chem. B. 2005, 109, 4707.

7. Chaparro, A. M.; Mueller, N.; Atienza, C.; Daza, L. J. Electroanal. Chem. 2006, 591, 69.

8. Vijayaraghavan, G.; Stevenson, K. J. Langmuir 2007, 23, 5279.

9. Jeon, N. L.; Lin, W.; Erhardt, M. K.; Girolami, G. S.; Nuzzo, R. G. Langmuir 1997, 13, 3833.

10. Crane, E. L.; You, Y.; Nuzzo, R. G.; Girolami, G. S. J. Am. Chem. Soc. 2000, 122, 3422.

11. Lin, W.; Warren, T. H.; Nuzzo, R. G.; Girolami, G. S. J. Am. Chem. Soc. 1993, 115, 11644.

12. Serp, P.; Kalck, P.; Feurer, R. Chem. Rev. 2002, 102, 3085.

13. Crane, E. L.; Girolami, G. S.; Nuzzo, R. G. Acc. Chem. Res. 2000, 33, 869.

14. Tagge, C. D.; Simpson, R. D.; Bergman, R. G.; Hosteler, M. J.; Girolami, G. S.; Nuzzo, R. G. J. Am. Chem. Soc. 1996, 118, 2634.

15. Xue, Z.; Strouse, M. J.; Shuh, D. K.; Knobler, C. B.; Kaesz, H. D.; Hicks, R. F.; Williams, R. S. J. Am. Chem. Soc. 1989, 111, 8779.

16. Xue, Z.; Thridandam, H.; Kaesz, H. D.; Hicks, R. F. Chem. Mater. 1992, 4, 162.

128

17. Norman, J. A. T.; Muratore, B. A.; Dwyer, D. N.; Roberts, D. A.; Hochberg, A. K. J. Phys. IV 1991, 1(C2), 271.

18. Powell, Q. H.; Kodas, T. T.; Anderson, B. M. Chem. Vap. Deposition 1996, 2, 179.

19. Zhu, Y.; Li, C.; Wu, Q. Surf. Coat. Technol. 2000, 135, 14.

20. Zhang, J.; Sasaki, K.; Sutter, E.; Adzic, R. R. Science, 2007, 315, 220.

21. Vahlas, C.; Juarez, F.; Feurer, R.; Serp, P.; Caussat, B. Chem. Vap. Deposition 2002, 8, 127.

22. Chen, Y. J.; Kaesz, H. D.; Thridandam, H.; Hicks, R. F. Appl. Phys. Lett. 1988, 53, 1591.

23. Cuesta, A.; Dhamelincourt, P.; Laureyns, J.; Martinez-Alonso, A.; Tascon, J. M. D. J. Mater. Chem. 1998, 8, 2875.

24. Papakonstantinou, P.; Lemoine, P. J. Phys-Condens. Matter 2001, 13, 2971.

25. Lo, S. H. Y.; Wang, Y-Y.; Wan, C-C. J. Colloid Interface Sci. 2007, 310, 190.

26. Serp, P.; Corrias, M.; Kalck, P. Appl. Catal. A. 2003, 253, 337.

27. Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl. Catal. B. 2005, 56, 9.

28. Solla-Gullon, J.; Rodes, A.; Montiel, V.; Aldaz, A.; Clavilier, J. J. Electroanal. Chem. 2003, 554, 273.

29. Grden, M.; Piascik, A.; Koczorowski, Z.; Czerwinski, A. J. Electroanal. Chem. 2002, 532, 35.

30. Pang, L. S. K.; Saxby, J. D.; Chatfield, S. P. J. Phys. Chem. 1993, 97, 6941.

31. Eley, D.D.; Pearson, E.J. J. Chem. Soc. Faraday I 1978, 74, 223.

32. Maillard, F.; Savinova, E. R.; Stimming, U. J. Electroanal. Chem. 2007, 599, 221.

129

33. Andreaus, B.; Maillard, F.; Kocylo, J.; Savinova, E.; Eikerling, M. J. Phys. Chem. B. 2006, 110, 21028.

34. Mc Callum, C.; Pletcher, D. J. Electroanal. Chem. 1976, 70, 277.

35. Arenz, M.; Mayrhofer, K.J.J.; Stamenkovic, V.; Blizanac, B.B.; Tomoyuki, T.; Ross, P.N.; Markovic, N.M. J. Am. Chem. Soc. 2005, 127, 6819.

36. Maldonado, S.; Stevenson, K. J. J. Phys. Chem. B. 2004, 108, 11375.

37. Climent, V.; Markovis, N. M.; Ross, P. N. J. Phys. Chem. 2000, 104, 3116.

38. Lukaszewski, M.; Czerwinski, A. J. Electroanal. Chem. 2006, 589, 38.

130

CHAPTER 5

Future Directions

5.1 ITRODUCTIO

One of the primary factors that prevents the commercialization of fuel

cells is kinetic limitations associated with the oxygen reduction reaction (ORR) at

the cathode.1 Fundamental understanding of the design, synthesis and utilization

of efficient catalysts is of vital importance in addressing these limitations. Current

research in the design of efficient catalysts for ORR primarily focuses on Pt based

multimetallic catalysts that work in a synergistic fashion. For example, recent

reports from the Adzic group2,3

that have compared carbon supported

monometallic Pt (Pt/C) and bimetallic PtAu (PtAu/C) have shown that the Pt/C

catalyst loses over 45 % of its electroactive surface area after 30,000 cycles in

oxygen saturated HClO4. Under the same experimental conditions the bimetallic

PtAu/C showed no change in the electroactive surface area. The stability of the

PtAu bimetallic catalyst was attributed to the decreased oxidation of the Pt

nanoparticles covered by the Au surface. Similar studies performed by Markovic

et al.4 on Pt3Ni (111) have shown that the particular bimetallic combination was

10 fold more active than a Pt (111) surface and 90 fold more active than currently

available commercial Pt/C catalysts for ORR. Strong ORR activity at the Pt3Ni

(111) surface was attributed to a highly favorable arrangement of the surface

atoms that promoted weak interactions between Pt surface atoms and non reactive

131

oxygenated species which increased the number of surface sites available for

oxygen adsorption and reduction. These fundamental studies serve as good

indicators for directions that need to be explored towards synthesis,

characterization and utilization of efficient ORR catalysts.

The previous three chapters were devoted to the fundamental

understanding of ORR at nitrogenated carbon nanotubes (NCNTs) and progress in

understanding ORR activity at metal organic chemical vapor deposition

(MOCVD) based and dendrimer templated catalysts supported on the NCNTs.

This chapter will focus on a three aspects of ongoing research, the first of

which is related to the efficient synthetic strategies to electrodeposit mono- and

multimetallic nanoparticle catalysts directly on NCNTs, second, research on

characterization protocols that enable understanding of the ORR process at

electrochemically dealloyed PtCu bimetallic catalysts supported on carbon that

are speculated to have undergone selective Cu dissolution and finally, preliminary

data from CNT supported catalysts utilized in a half cell ORR process that

simulated fuel cell conditions. These experiments were carried out in

collaboration with members of the Stevenson and Manthiram research groups.

132


5.2.1 Electrodeposition of anoparticle Catalysts on CTs

The concept of electrochemical deposition of nanoparticle catalysts has

been explored by a few research groups recently. Notable schemes include

nanoparticles deposited electrochemically on glassy carbon (GC) substrates,5

nanoparticles deposited on CNTs grown on carbon cloth6 and nanoparticles

deposited on CNTs cast on GC substrates.7

Some of the problems associated with electrodeposition on GC involve

irregular dispersion of the nanoparticles on the GC surface and poor control of

nucleation and growth of the nanoparticle due to difficulties in defining surface

sites on the GC.5 Difficulties in the electrodeposition of catalysts on CNTs have

much in common with difficulties at the GC surfaces but can be overcome with

aggressive preprocessing steps that create surface functionalities on the CNTs

allowing for better anchoring, nucleation and dispersion of nanoparticles.8 The

drawback to using these preprocessing steps is that they degrade structure and

composition of the CNT, leading to broad variations in stability.

The use of NCNTs is an attractive option for the synthesis and

characterization of nanoparticles through electrodeposition in the face of the

difficulties described above. As discussed in the previous chapters, properties of

NCNTs can be controlled to a remarkable degree. Systematic manipulation of the

edge plane sites and surface concentration of nitrogen facilitate controlled

133

nucleation, dispersion and growth of nanoparticles that make active carbon/

catalyst composites for ORR.

Studies involving electrodeposition of Au and PtAu nanoparticles on

NCNTs are primarily detailed in this section. These studies were conducted in

collaboration with Mr. Jay Sawyer Croley at the Stevenson research lab. The

NCNTs used for electrodeposition were grown directly on a nickel mesh substrate

cut to specific dimensions so as to normalize the surface area as described in the

experimental section in chapter 2.

Au nanoparticles were deposited on 4 % NCNTs from a solution of 0.2

mM HAuCl4. A step in potential from open circuit to -0.01 V vs. NHE for 0.1 ms

facilitated the electrodeposition. PtAu bimetallic nanoparticles were synthesized

in a subsequent step after the deposition of Au at -0.1 V for 0.1 ms. The

underpotential electrodeposition of Au on Pt is currently being explored. Figure

5.1 shows a representative chronocoulometric curve from the electrodeposition of

Au nanoparticles on 4 % NCNT supports. The systematic study of the charge

associated with the deposition of nanoparticles with respect to time provides a

guideline that facilitates effective control of the nanoparticle size and

composition.

Figure 5.2 shows representative TEM images of nanoparticles

electrodeposited on 4 % NCNTs. Figure 5.2a shows that the Au nanoparticles are

nearly spherical and uniformly dispersed on the NCNT surface. Edge plane sites

at the NCNT surface are speculated to play a role in the uniform dispersion of the

Au particles as evident by an absence of particle agglomeration. PtAu bimetallic

134

0 2 4 6 8 102.5

2.0

1.5

1.0

0.5

0.0

Charge / 1e-7 C

Time / 1e-5 s

Figure 5.1 Chronocoulometric profile of the electrodeposition of Au nanoparticles from 0.2 mM HAuCl4 solutions on NCNTs

135

Figure 5.2 Representative TEM images of nanoparticle catalysts deposited on 4 % NCNTs. a) Au nanoparticles. b) PtAu nanoparticles.

b)

a)

136

particles shown in Figure 5.2b were found to be similarly well dispersed. A high

resolution image of the PtAu nanoparticles in Figure 5.3 shows that the as

deposited nanoparticles were crystalline.

The activity for ORR at the PtAu/ NCNT composites was tested in oxygen

saturated 1 M KNO3 since the NCNTs were grown on a Ni mesh substrate. Figure

5.4 shows representative cyclic voltammograms for ORR at PtAu nanoparticles

supported on 4 % NCNTs. A 200 mV positive shift in the peak potential for ORR

was seen at the 4 % NCNT/ PtAu composites, showing that they were active for

ORR. Further studies in order to provide better control of the particle size and

composition of these bimetallic catalysts are underway.

Fundamental studies on the synthesis of mono- and multimetallic

nanoparticles on NCNTs through electrodeposition provide insight into direct

synthetic strategies for producing active ORR catalysts. These studies also

provide solutions that circumvent a majority of pre- and post processing steps

required for synthesis of traditional carbon supported catalysts. For example the

Adzic group used galvanostatic displacement to deposit Au on Pt which showed

remarkable enhancement in ORR activity. The use of NCNTs in the

electrodeposition process enables synthesis of varying compositions of AuPt

bimetallic catalysts without the need for galvanostatic displacement. The

electrodeposition of well defined nanoparticles without the need for expensive

templating agents also emphasizes the cost effective nature of this strategy.

137

Figure 5.3 High resolution TEM image of PtAu nanoparticles electrodeposited on NCNTs.

138

Figure 5.4 Cyclic voltammograms for oxygen reduction at PtAu catalysts deposited on NCNTs. CV’s were run in oxygen saturated 1 M KNO3.

0.6 0.4 0.2 0.0 -0.2

-0.6

-0.3

0.0

0.3

0.6

0.9

Current / mA


Control 6.5 % NCNTs

PtAu on 6.5 % NCNTs

139

5.2.2 Synthesis and Characterization of ORR at CT supported PtCu Catalysts

Selective electrodissolution of Cu atoms from PtCu bimetallic

nanoparticle catalysts has been a topic of interest recently due to significant

enhancement in the ORR activity at the resultant catalyst. It has been speculated

that the selective electrochemical dealloying of the surface Cu atoms creates

preferred structural rearrangement of the Pt atoms.9,10

The structural

rearrangements are hypothesized to include active crystal facets or favorable Pt-Pt

interatomic distances that enhance the adsorption and reduction of oxygen at these

catalyst surfaces.

While selective dissolution studies clearly show significant enhancement

in the catalytic activity for ORR at carbon supported PtCu9 and PtCuCo

10

catalysts, the mechanistic details still remain topics of debate. Although it has

been hypothesized that dealloying of Cu does not increase the surface area of the

resultant catalyst, characterization steps performed on those specific catalysts lack

convincing proof to support those specific claims.

In an effort to improve upon the existing characterization data that try to

explain enhanced catalytic activity at dealloyed surfaces, a set of basic

measurements were made. These involved i) understanding dealloying aspects at

Au surfaces through repeated alloying/ dealloying cycles with Zn. ii) synthesis of

PtCu nanoparticle catalysts using dendrimer templates (as described in chapter 3)

and testing for ORR activity at dealloyed PtCu dendrimer encapsulated

nanoparticles (DENs) supported on NCNTs. iii) the development of new

140

characterization strategies that enable interrogation of the PtCu catalyst surface

before and after dealloying.

5.2.2.1 Fundamental Studies on Dealloying at Au Thin Films

Figure 5.5 shows an SEM image of a Au thin film after 30 alloying/

dealloying cycles with Zn. Au films were deposited on an Indium tin oxide (ITO)

substrate. Electrochemical alloying/ dealloying was carried out in a standard three

electrode electrochemical cell with the Au film serving as the working electrode,

a thin zinc strip serving as a reference and another piece of zinc serving as a

counter electrode. 1.6 M ZnCl2 in benzyl alcohol was used as the electrolyte and

the working electrode was cycled between -0.7 and 1.8 V vs. Zn.11

The resultant

film was found to have a roughened surface and exhibited a presence of pores

created by the alloy/ dealloy process. The Au film also had a few spots on the

surface where the Au had been removed from the ITO surface due to the cycling.

The difference in the electrochemical surface area for the Au film before

and after the alloy/ dealloy cycling was measured by cyclic voltammetry of the

film in 0.5 M H2SO4. The Au film served as the working electrode while a Pt wire

and an Hg/ Hg2SO4 electrode were used as counter and reference electrodes,

respectively. Potentials are reported vs. NHE for comparison. Cyclic

voltammograms of the Au films before and after Zn alloying/dealloying are

shown in Figure 5.6. Electrochemical surface area of the Au film was determined

by integration of the charge associated with the reduction of AuO during the

cathodic sweep of the voltammogram. An 17.1 % increase in the electrochemical

141

Figure 5.5 Scanning electron microscope image of Au film subjected to 30 alloy/ dealloy cycles in a ZnCl2 / benzyl alcohol solution.

142

0.0 0.2 0.4 0.6 0.8 1.0-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

Current / mA

Potential / V vs. Zn

Pristine Au film

Au film after 30 AuZn cycles

Figure 5.6 Cyclic voltammograms of an Au film in 0.5 M H2SO4 before and after alloying/ dealloying cycles in a ZnCl2 / benzyl alcohol solution. Integration of the cathodic AuO peak showed a 17.1 % increase in the electrochemical surface area on the dealloyed Au film.

143

surface area was seen at the Au film after 30 alloying/ dealloying cycles with Zn.

While it is clear that there is an increase in the electrochemical surface area it

would be of interest from a fundamental standpoint to determine if there was an

increase in the physical surface area of the resultant Au film and how the

geometric surface area changes with respect to the electrochemical surface area.

Studies involving spectroscopic ellipsometry in conjunction with a custom made

cell that would enable toluene dosing at the Au films are planned to elucidate

differences in the physical surface area as a result of alloying/ dealloying steps.

These fundamental studies are expected to improve understanding of the ORR

process at dealloyed nanoparticle catalysts.

5.2.2.2 Effects of Electrochemical Dealloying on ORR at PtCu DE!s supported on 6.5 % !C!Ts.

Preliminary studies were conducted to determine the effects of

electrochemical dealloying for ORR activity at PtCu nanoparticle catalysts. PtCu

DENs were prepared with assistance of Ms. Wenly Ruan. Briefly, generation 4,

amine terminated dendrimers were used to coordinate with Pt and Cu ions from

solution and then reduced with NaBH4 to form 1.4 nm PtCu bimetallic

nanoparticles. The as synthesized PtCu DENs were adsorbed onto 6.5 % NCNTs

by suspending the NCNTs in the DEN solution and stirring for 12 h similar to

protocols described in chapter 3.

Selective Cu dissolution from the PtCu DENs was carried out in 0.1 M

HClO4. A sample slurry of a measured mass of the PtCu DENs supported on

NCNTs in ethanol was drop cast on a GC working electrode suspended in a three

144

-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

Current / mA


PtCu DEN / 6.5 % NCNT Cycles 1-5




Figure 5.7 Electrochemical dissolution of Cu from PtAu bimetallic catalysts synthesized using dendrimer templates. The PtCu/ DEN catalysts were supported on 6.5 % NCNTs. Dissolution studies were carried out in 0.1 M HClO4.

145

electrode cell with a gold counter and Hg/ Hg2SO4 reference. Potentials are

reported vs. NHE for comparison. Figure 5.7 shows dealloying cycles 1-20 at the

PtCu DENs. Features on the CV’s related to the alloying and dealloying of Zn

with Au compare well with literature reports on similar systems. Current densities

of the peak at 0.4 V associated with the Cu decrease with the number of

dealloying cycles due to the oxidation of Cu from the catalyst surface. More

significantly, features associated with the adsorption/ desorption of hydrogen (0 -

0.2 V) at the Pt surface seem to increase in current density due to a larger number

of Pt atoms being exposed on the surface for hydrogen adsorption/ desorption.

Fundamental studies involving varied compositions of PtCu are currently being

carried out to elucidate ORR activity at dealloyed surfaces.

5.2.2.3 Devising Improved Synthesis and Characterization Strategies for ORR Catalysts

While a large number of reports have delved on ORR over the past few

decades, the need for effective and cost efficient catalysts still exists. Most reports

have discussed empirical and applied aspects of catalysts and fuel cells but have

not conducted fundamental studies that understand ORR at catalyst composites

through advanced characterization strategies. Towards this end, some of the

research at the Stevenson research lab has been focused on devising hybrid

characterization strategies that elucidate the state of a nanoparticle catalyst before

and after it is subject to ORR.

Some of the recent efforts involve the growth of NCNTs directly on

customized Ni and Au TEM grids. Nanoparticle catalysts can then be

incorporated on the NCNT surface either by suspending the grid in a DEN

146

solution or by electrochemical means – similar to protocols discussed in chapter 3

and section 5.2.1. The NCNT supported catalysts might be characterized at the

TEM, subject to ORR in acidic or basic media and then characterized again using

the TEM. These measurements enable direct analysis of the nanoparticle surface

using TEM and energy dispersive spectroscopy (EDS) before and after ORR. The

size, structure and composition of the catalyst before and after ORR can be

directly correlated to the corresponding activity enabling significantly better

understanding of the catalysis, thereby helping design efficient catalysts.

5.2.3 CT Supported Catalysts in Fuel Cells

CNT supported ORR catalysts were tested in a half cell configuration for

oxygen reduction under simulated fuel cell conditions. These studies were

conducted in collaboration with Dr. Raghuveer Vadari at the Manthiram research

group. Pt DEN catalysts were adsorbed on CNTs resulting in CNT/ Pt-DEN

composites that were spray coated on a carbon cloth from a suspension in nafion.

A 1 cm2 area of the thus coated electrode containing 0.05 mg / cm

2 Pt loading was

tested for ORR activity in 1 M H2SO4.

Figure 5.8 shows a representative half cell ORR process at CNT/ Pt-DENs

compared to commercial 20 wt. % Pt on Vulcan XC-72. While the current

densities are lower for the CNT/ Pt-DENs, it should be noted that the catalyst

loading was lower than 50 % than that at the commercial catalyst. Such studies

show promise for CNT supported catalysts and further studies involving NCNT

supported catalysts are planned for the future.

147

Figure 5.8 Half cell trial of CNT supported Pt DEN catalysts for ORR in 1 M H2SO4.

0.00 -0.05 -0.10 -0.15 -0.200.0

0.2

0.4

0.6

0.8

1.0 Commercial Pt

CNT/ Pt-DEN


Current density / A cm-2

148

5.3 COCLUSIOS

Efficient synthetic strategies for cost effective and active ORR catalysts

have been explored through electrodeposition. Preliminary data show promise for

extending these studies to different catalyst compositions that minimize the

amount of Pt loading while increasing ORR activity.

Studies on selective electrochemical dealloying of bimetallic catalysts

were carried out in effort to elucidate the effects of this process on ORR activity.

Auxiliary studies on alloying/ dealloying at Au surfaces and studies on PtCu

alloys synthesized using dendrimer templates and supported on NCNTs aim to

detail changes in the catalyst surface structure, surface area and composition that

correlate to enhanced ORR activity.

Preliminary half cell ORR studies on CNT supported Pt-DEN catalysts

show promise for carbon nanotube supported catalysts in fuel cells.

149

5.4 REFERECES

1. Gasteiger, H. A.; Kocha, S. S.; Sompalli, B.; Wagner, F. T. Appl. Catal. B

2005, 56, 9.

2. Zhang, J.; Sasaki, K.; Sutter, E.; Adzic, R. R. Science, 2007, 315, 220.

3. Strbac, S.; Adzic, R. R. J. Electroanal. Chem. 1996, 403, 169.

4. Stamenkovic, V. R.; Fowler, B.; Mun, B. S.; Wang, G.; Ross, P. N.;

Lucas, C. A.; Markovic, N. M. Science, 2007, 315, 493.

5. Papadimitriou, S.; Tegou, A.; Pavlidou, E.; Kokkinidis, G.; Sotiropoulos,

S. Electrochimica Acta, 2007, 52, 6254.

6. Tsai, M. –C.; Yeh, T. –K.; Tsai, C. –H. Electrochem. Comm. 2006, 8,

1445.

7. Zhao, Y.; Fan, L.; Zhong, H.; Li, Y.; Yang, S. Adv. Funct. Mater. 2007,

17, 1537.

8. Xu, Y.; Lin, X. Electrochimica Acta 2007, 52, 5140.

9. Koh, S.; Strasser, P. J. Am. Chem. Soc. 2007, 129, 12624.

10. Srivastava, R.; Mani, P.; Hahn, N.; Strasser, P. Angew. Chem. Int. Ed.

2007, 46, 1.

11. Jia, F.; Yu, C.; Ai, Z.; Zhang, L. Chem. Mater. 2007, 19, 3648.

150

Vita

Ganesh Vijayaraghavan, the son of Saraswathi and Vijayaraghavan

Velayutham was born in Madurai, India in June 1978. After graduating from high

school, he pursued a Bachelor’s degree in Chemical and Electrochemical

Engineering at the Central Electrochemical Research Institute in Karaikudi, India.

After completion of a Master’s degree in Analytical Chemistry from Texas Tech

University in 2004, he joined the University of Texas at Austin to pursue his

Doctorate in Analytical Chemistry under the guidance of Prof. Keith J. Stevenson.

Permanent address: 3405 Helms St #104, Austin TX 78705

This dissertation was typed by the author.

copyright by ganesh vijayaraghavan 2007

Documents